NICKEL DEPOSITION ON HYDRODEMETALLATION CATALYSTS
XINJIN ZHAO
B.S., Taiyuan University of Technology (1982) M.S., Institute of Coal Chemistry, Academia Sinica (1986) M.S.C.E.P., Massachusetts Institute of Technology (1990) Submitted to the Department of Chemical Engineering in partial fulfillment of the requirements for the degree of Doctor of Science in Chemical Engineering at the MASSACHUSETTS INSTITUTE OF TECHNOLOGY
February 1993 ®Massachusetts Institute of Technology 1993. All rights reserved.
Author
Department of Chemical Engineering November 13, 1993
Certified by James Wei Professor E/ritus of Chemical Engineering Thess Supervisor
Accepted by Robert E. Cohen Chairman, Departmental Committee on Graduate Students
MASSACHUSETTSINSTITUTE OFTECHNOLOGY
FEB 26 1993 ARCHIVES LBRARIES NICKEL DEPOSITION ON HYDRODEMETALLATION CATALYSTS by Xinjin Zhao
Submitted to the Department of Chemical Engineering on November 13, 1992, in partial fulfillment of the requirements for the degree of Doctor of Science in Chemical Engineering
Abstract The prevailing metals in petroleum are nickel and vanadium which exist in porphyrins and asphaltenes. These organometallic molecules are large and their sizes approach the pore sizes of hydrotreating catalysts. As a result, these compounds deposit on the catalyst surface during hydrotreating processes and irreversibly destroy the cat- alysts by affecting transport into intraparticle fine pores as well as causing chemical changes when the deposited metals interact with the original active components on the catalysts. A better understanding of the deposition phenomena on the catalysts would establish a basis for developing improved hydrotreating catalysts. Nickel etio-porphyrin was used as a model compound to study nickel deposition and the interaction of the deposits with the catalytic components on presulfided CoO - MoO3/A1 203 hydrodemetallation catalysts under industrially-relevant pro- cess conditions, though no diffusion effects were present in the study. The structures of the nickel deposits on the aged catalysts were characterized by various microscopic and spectroscopic techniques. The nickel deposits were identified as nickel sulfide (Ni7S 6) in crystallite form. At a nickel loading of about 20%, the average size of the crystallites was estimated to be about 10 to 15 nanometers, while crystallites with the sizes up to 100 nanometers were also observed. X-ray elemental mapping and X-ray microanalysis on a dedicated scanning trans- mission electron microscope (STEM) and high resolution transmission electron mi- croscope (HRTEM) showed that nickel sulfide deposits were strongly associated with cobalt sulfide (Co9 s8) on the catalyst. In contrast, only about 20 to 25% of the molybdenum was associated with the nickel deposits as a segregated molybdenum disulfide (MoS 2) surface layer phase, the rest of the molybdenum disulfide forms 3
separate entities on the catalyst surface. The association between cobalt and nickel sulfides was shown to be a result of solid solution formation between the two sulfides, while the segregation of molybdenum sulfide is due to its lower surface energy. Segregation of molybdenum disulfide was quantitatively determined by X-ray microanalysis on scanning transmission electron microscope and directly observed on high resolution transmission electron microscope. It was also showed that the degree of segregation decreases for crystallites smaller than about 15 nanometers. Nickel sulfide deposits enhance the sintering of the catalytic metal sulfides by low- ering their Tammann temperatures. Electron microscopic studies showed that the sintering of the catalytic metal sulfides, especially cobalt sulfide, increased with the presence of nickel sulfide deposits. The effect was also discussed on hypothetical phase diagrams. It was showed that there existed a threshold nickel loading above which the catalytic metal sulfides would become mobile. The threshold value was depen- dent on the specific system and the operating temperature. The enhanced sintering of molybdenum sulfide, rather than covering-up of active sites by deposits as being suggested in many literatures, was considered as the major cause for deactivation by metal deposits at diffusion-free conditions. The morphology of the metal deposits has significant impact on the deactivation of hydrodemetallation catalyst. Therefore, our results could have important implica- tion. Since segregated large crystallites of nickel deposits lead to less deactivation, in comparison with small crystallites or uniform layer deposition, improved hydrodemet- allation catalyst could be developed by manipulating the morphology of cobalt on the catalyst surface.
Thesis Supervisor: James Wei
Title: Professor Emeritus of Chemical Engineering Massachusetts Institute of Technology
Dean School of Applied Science and Engineering Princeton University Acknowledgments
I am indeed indebted to Professor James Wei for mentorship, for lighting the way, for his technical and personal guidance, for an environment I could grow as a scientist. I am also very grateful to my thesis committee members, Professor Charles N. Satterfield, Dr. Miretta F. Stephanopoulos, Professor Klavs F. Jensen and Dr. An- thony J. Garratt-Reed for sharing their wealth of knowledge and many insights during the course of the investigation. Dr. Garratt-Reed deserves a special mention for his invaluable assistance in performing STEM analysis. I would also like to thank Mr. Michael Frongillo for the superb high resolution TEM work. The financial support for the research from Mobil Oil Research and Development Corporation and Chevron Research Company, as well as the Research Assistantship and Fellowships from Department of Chemical Engineering of Massachusetts Institute of Technology are gratefully acknowledged. I also thank Dr. James D. Carruthers and Dr. Robert H. Whitman of American Cyanamid Company at Stamford, Conneticut for providing catalyst samples. I wish to acknowledge the valuable discussions and various supports from the past members of the hydrodemetallation laboratory, especially Dr. Barbara Smith, and Dr. Chi-Wen Hung at Chevron Research Company, and Dr. Kirk Limbach at Rohm and Hass Company. During the first stage of the thesis, fruitful discussions and advice from them were most helpful in defining the scope of the thesis. The completion of the thesis has cost me four years, but they have not been devoid of excitement. The many friends with whom I shared the time have made of MIT a joyful memory, knowing them has been a special prize of four years at MIT. Be it a Friday basketball game, an overnight Practice School visit to Tahoe Casino, or a weekend trip to Cape Cod beach, we always managed to have a good time and still get our work done. Grateful appreciation is expressed to the following, though the 5
list is far from complete: Leo Lue, for all those fun time even since you came to the Zoo; Joy Mendoza, for all the joys we had on the basketball court; Gordon Smith, for the fish and lobsters you cooked; Stathis Avgoustiniatos, for all the good and bad time we shared at Practice School; Marc Moran, for teaching me from juggling to driving; and the many others who were responsible, directly or indirectly, wittingly or unconsciously, to my being and well-beings; ..... I benefited greatly from the friendship and advice of Jirong Xiao, Jiang Yue, Zhicheng Hu. Their helps are greatly appreciated. Many thanks go to Hojoon Park, my first-year roommate, who helped me to adjust from the eastern to the western culture in my crucial first term, and told me the difference between spaghetti and noodle. I wish you the best. Thanks are also due to many other friends who contributed my education at MIT. Of those, Yaping Liu for his help in TEM, and Qing Huang for first introducing me into the world of Athena, where the thesis is eventually born'. Our secretary, Linda Mousseau catered to my every need in a most generous and cheerful way, for which I am very appreciative. My deepest appreciation and love go to my wife, Luhong and my far-away parents, who have sacrificed so much for me, and whose unwavering support and love I have depended on my life. Without them, I could never have taken this endeavor. Thank you, Luhong, Thank you, Mom and Dad, I love you all. Always, in all ways.
1The thesis is written with IATEX. To Luhong Who spent so many lonely nights and weekends during the last four years, but wholeheartedly supported my pursuit with love, encouragement and understanding.
To My Parents who never had a chance to have a formal education, but fully understand the importance and benefits of educat- ing their children to the best of their capabilities. Contents
Abstract 2
Acknowledgments 4
List of Figures 12
List of Tables 15
1 Introduction and Objectives 16
1.1 Background ...... 16
1.2 Motivation and Objectives . . 18 1.3 Literature Review ...... 20
1.3.1 Structure of Sulfided Co - Mo/ - A120 3 Catalyst . . . . . 20 1.3.2 Deposition Patterns. . 24
1.3.3 Catalyst Deactivation...... 27
1.3.4 Migration of Metals on Catalyst Surface ...... 29
2 Thermodynamic Considerations 31
2.1 Introduction...... 31
2.2 Phase Diagrams ...... 32
2.2.1 Systems with Hydrogen and Hydrogen Sulfide ...... 34
7 CONTENTS 8
2.3 Surface Segregation ...... 34
3 Hydrodemetallation Experiments 41
3.1 Chapter summary ...... 41
3.2 Introduction ...... 42 3.3Equipment ...... 42 3.4 Model Compounds ...... 43
3.5 Solvent ...... 43 3.6 Catalysts...... 47 3.7 Hydrodemetallation ...... 47
3.8 Characterization ...... 52
3.8.1 High Resolution Transmission Electron Microscope (HRTEM) 58
3.8.2 Scanning Transmission Electron Microscope (STEM). 58
3.8.3 X-ray Photoelectron Spectroscopy (XPS) . 60
3.8.4 X-ray Diffraction Analyzer (XRD) ...... 61
3.8.5 Surface Area Measurement(BET). 61
4 Nickel Deposition on Co - Mo Catalyst 62 4.1 Chapter Summary ...... 62 4.2 Introduction ...... 63 4.3 Electron Microscopy Results ...... 63 4.3.1 Bare Catalyst ...... 64 4.3.2 Sulfided Catalysts ...... 68 4.3.3 Unsulfided Catalysts ...... 72 . . . . . 4.4 X-ray Diffraction Results ...... 75 . . . . . 4.5 Discussions ...... 75 4.6 Conclusions ...... 80 CON~TENTS 9 9 CONTENTS 5 Deposition Mechanism 82
5.1 Chapter Summary ...... 82 5.2 Introduction...... 83
5.3 Migration Experiment ...... r 85
5.4 Characterization Results ...... 86
5.4.1 Catalyst with Impregnated Nickel ...... 86 5.4.2 Catalyst with Impregnated Nickel after Being Treated 86 5.5 Discussion ...... 87 5.5.1 Tammann Temperature ...... 87 5.5.2 Surface Diffusivity ...... 92 5.5.3 Affinity between Nickel and Cobalt Sulfides ...... 92
5.6 Conclusions ...... 96
6 Metal Distribution within Deposition C.rystallites 99 6.1 Chapter Summary ...... 99 6.2 Introduction...... 100 6.3 Characterizations ...... 100 6.3.1 Characterization by STEM ...... 100 6.3.2 Characterization by HRTEM ...... 113 6.3.3 Characterization by XPS ...... 118 6.4 Discussions ...... 120 6.5 Conclusions ...... 123
7 Mobility of Catalytic Metal Sulfides on Catalyst Surface 125 7.1 Chapter Summary ...... 125...
7.2 Introduction...... 126... 7.3 Deactivation of Catalysts by Sintering ...... 126... 7.4 Mobility in Two Component Systems 129 CONVTENTS 10 10 CONTENTS 7.5 Hydrodemetallation Catalyst Surface ..... 133 7.6 Electron Microscopic Results ...... 137 7.6.1 Sulfided Bare Catalyst ...... 137 7.6.2 Heat Aged Catalyst ...... 137 7.6.3 Nickel Aged Catalyst ...... 138 7.7 Conclusions ...... 140
8 Nickel Deposition and Catalyst Deactivation 143 8.1 Chapter Summary ...... 143 8.2 Development of Nickel Deposits ...... 144 8.3 Catalyst Deactivation...... 145 8.4 Approaches for Improved Catalyst Design 148 8.5 Conclusions. 149
9 Conclusions 153
10 Recommendations 156
Appendices 159
A STEM analysis data 159 A.1 XEDS Microanalysis: Unsulfided HDS16A ...... 160
A.2 XEDS Microanalysis: Sulfided HDS16A ...... 161
A.3 XEDS Microanalysis: HDS16A with Impregnated Nickel ...... 162 A.4 XEDS Microanalysis: HDS16A with Impregnated Nickel after Heating 163
A.5 XEDS Microanalysis: HDS16A (Within One Crystallite) ...... 164 A.6 XEDS Microanalysis: SN6931 (Within One Crystallite) ...... 165 A.7 XEDS Microanalysis: Surface Segregation Data ...... 166
B Acronyms 167 CONTENTS 11
Bibliography 168 List of Figures
1-1 General Representation of CoO - MoO3/l - A120 3 Catalyst ..... 22 1-2 Deactivation of Hydrotreating Catalyst in Pilot-Plant Experiments[34] 28
2-1 Phase Diagram of Ni-Co-S at 1273K[40] 33 2-2 Hydrogen Reduction of Nickel Sulfides[63] ...... 35 2-3 Hydrogen Reduction of Cobalt Sulfides[63] ...... 36
2-4 Hydrogen Reduction of Molybdenum Sulfides[63] . . . 37
2-5 Possible Microstructures of a Highly Dispersed Alloy in a Substrate 38
3-1 Schematic of the Hydrodemetallation Equipment . . . 44 3-2 Schematic of the l-litre Autoclave Reactor ...... 45 3-3 Molecular Structure of Nickel Etio-Porphyrin ...... 46 3-4 Schematic Diagrams of Ultramicrotomy ...... 57 3-5 Signals Created by the Interaction of High Energy Elecitrons with the Specimen. 59
4-1 Electron Micrograph of Bare HDS16A Catalyst 65 4-2 Elemental Mapping of Bare HDS16A Catalyst . . 66 4-3 High Resolution Image of Sulfided Bare HDS16A Catalyst 67 4-4 Electron Micrograph of Aged Sulfided SN6931 Catalyst . . 69 4-5 Elemental Mapping of Aged SN6931 Catalyst with Sulfur . 70
12 LST OF FIGURES 13 LIST OF FIGURES 13~~~__ 4-6 Elemental Mapping of Aged HDS16A Catalyst with Sulfur .... . 71 4-7 EDS Microanalysis of Aged HDS16A Catalyst ...... 73 4-8 EDS Microanalysis of Aged HDS16A Catalyst ...... 74 4-9 Elemental Mapping of Aged HDS16A Catalyst without Sulfur ... . 76 4-10 EDS Microanalysis of Aged HDS16A Catalyst without Sulfur .... . 77 4-11 EDS Microanalysis of Aged HDS16A Catalyst without Sulfur ..... 78 4-12 X-ray Diffraction Spectra of Aged HDS16A Catalyst ...... 79
5-1 Activity of Catalysts with different Cobalt Contents[9] ...... 84 5-2 Elemental Mapping of HDS16A with Impregnated Nickel ...... 88 5-3 Microanalysis of HDS16A with Impregnated Nickel ...... 89 5-4 Elemental Mapping of HDS16A with Impregnated Nickel after Treating 90 5-5 Microanalysis of HDS16A with Impregnated Nickel after Treating . . 91
6-1 Illustration of the Electron Probe on a Crystallite ...... 102 6-2 Electron Micrograph of a Crystallite Analyzed ...... 103 6-3 Element Ratios within the Crystallite Analyzed ...... 105 6-4 Element Distribution within the Crystallite Analyzed ...... 106 6-5 Electron Micrograph of a Crystallite Analyzed ...... 107 6-6 Element Ratios within the Crystallite Analyzed ...... 108 6-7 Element Distribution within the Crystallite Analyzed ...... 109 6-8 Sketches of Seven Crystallites Analyzed ...... 110 6-9 Molybdenum/Nickel Radial Distribution within Crystallites ...... 111 6-10 Cobalt/Nickel Radial Distribution within Crystallites ...... 112
6-11 Effect of Crystallite Sizes on Segregation ...... 114 6-12 Lattice Fringe Image of Aged HDS16 Catalyst ...... 116 6-13 Lattice Fringe Image of Aged HDS16 Catalyst ...... 117
6-14 Electron Micrograph of a Molybdenum Sulfide Crystallite . 119 LIST OF FIGURES 14 LISTOFFIGURES 14~~~~~~~~~~~~~_
6-15 Molecular Structure of MoS 2 [105] ...... 122
7-1 Effect of Absorbate Coverage on the Activation Energy of Tungsten Surface Self-Diffusion ...... 128 7-2 Mobility Regions on Hypothetical Phase Diagram: I. Complete Soluble System ...... 130 7-3 Mobility Regions on Hypothetical Phase Diagram: II. Partially Soluble System ...... 131 7-4 Mobility Regions on Hypothetical Phase Diagram: III. Complete In- soluble System ...... 132
7-5 Mobility of Co9 S8 with Ni S67 Deposits ...... 135
7-6 Mobility of MoS 2 with Ni 7S6 Deposits ...... 136 7-7 High Resolution Micrograph of Heat Aged Catalyst ...... 139
7-8 High Resolution MoS 2 Image on Aged Sulfided Catalyst ...... 141 7-9 Elemental Mapping of Aged HDS16 Catalyst ...... 142
8-1 Nickel Deposition Mechanism ...... 146 8-2 Hydrodesulfurization Activity with Different Promoter Contents in Catalysts [35] ...... 150 8-3 Hydrodemetallation Activity with Different Promoter Contents in Cat- alysts [35] ...... 151 List of Tables
1.1 Research Objectives ...... 21
3.1 Properties of Squalane...... 48 3.2 Compositions of Catalysts ...... 49 3.3 Properties of Catalyst HDS16A ...... 50 3.4 Summary of Hydrodemetallation Runs ...... 53 3.5 Ladd Ultra-Low Viscosity Embedding Medium ...... 55
5.1 Characteristic Temperatures of Metal Sulfides ...... 93 5.2 Field Strength of Metal Sulfides ...... 95 5.3 Structures and Properties of Metal Sulfides ...... 97
15 Chapter 1
Introduction and Objectives
Dr. Watson would tell you that these little digressionsof mine sometimes prove in the end have some bearing on the matter. -- Sherlock Holmes, The Adventure of the Three Garridebs Sir Arthur Conan Doyle
1.1 Background
Due to both the concern for the environment and the decreasing availability of oil, ever increasing quantities of crude oils and residuals have to be processed. One of the most important features of these crude oil and residuals is their high heteroatom contents. In addition to the improvement of carbon/hydrogen ratios, a major goal of hydrotreating is the removal of heteroatoms, which includes sulfur, nitrogen, oxygen, and trace metals. While the role of catalyst in light hydrotreating is mainly to promote selective removal of sulfur and nitrogen, the catalyst must additionally promote the
16 Introduction and Objectives 17
removal of metals in processing heavier feeds. Catalytic hydrotreating is usually performed in the presence of well established
catalyst system consisting of y - A120 3 supported combination of molybdenum or tungsten and cobalt or nickel, at elevated pressures and temperatures. In crude oils, nearly half of the metallic elements in the periodic table have been identified as trace elements [130]. The most abundant and troublesome metals in crude oil are nickel and vanadium, present in amounts ranging from a few ppm to over 1000 ppm. These metals usually exists in organometallic molecules, typically metal porphyrins and asphaltenes. Unlike other heteroatoms, such as sulfur, nitro- gen or oxygen, which can be removed as gaseous products after hydrotreating, metals stay and accumulate on the hydrotreating catalysts as metal sulfide deposits. The de- posits lead to irreversible catalyst deactivation which is a major problem in residuum hydrotreating and can often result in expensive catalyst replacement [85]. Successful ways to regenerate the spent catalysts with metal deposits are yet to be developed
[117]. The organometallic molecules are large and approach the same order of magnitude as the pore size of the hydrotreating catalysts. As a result, these compounds deposit close to the mouth of the pore after hydrogenolysis reactions. The deposits destroy the catalyst by affecting transport into intraparticle fine pores as well as causing chemical changes that occur when the deposited metal interacts with the original active sites on the catalyst [9] [19] [69] [113]. Although the diffusion within hydrodemetallation catalysts has been a subject of many experimental and modeling studies[29] [54][55] [60] [64] [75], the interaction of metal deposition with catalytic metals has received less attention [50] [113] [127]. Many technical advances are still based on empirical considerations. A better understanding of the chemical nature of the metal deposits and their interaction with the catalytic metals would establish a fundamental basis for developping improved hydroprocessing catalysts and reactors. Introduction and Objectives 18 Introduction and Objectives 18 In general, the characteristics of aged and spent catalysts have not been well defined though the metal deposits are increasingly found to take place as crystallites over discrete sites, both in laboratory conditions with clean oil and model compounds [71] [107] and in industrial pilot plant [116], which is in contrary to previous uniform layer assumptions[9] [19] [86]. The implications and control of the metal deposition would have significant consequences on hydrotreating catalyst design.
1.2 Motivation and Objectives
Catalytic hydrotreating of residuum oil is currently conducted at 1.2 million barrels a day in the world, at a replacement cost of more than 200 million dollars per year[126]. This cost is certainly going to increase sharply, as world concern for the environment continues to exert pressure on cleaner fuels. In addition, catalytic hydrotreating is also necessary for making usable coal liquefaction products. The necessity of the costly replacement of hydrodemetallation catalyst in industry has motivated fundamental study to extend the life span of hydrodemetallation catalysts. The ultimate goals for this studies are:
* To find efficient ways of using hydrodemetallation catalysts, and to extend the life span of the catalysts;
* To improve the design of hydrodemetallation catalysts.
The objective of this work is to investigate the governing factors that determine the deposition patterns of nickel and the interaction of the nickel deposits with cobalt and molybdenum on hydrodemetallation catalysts under diffusion-free conditions. With this information, we will further try to propose some approaches to hydrodemetalla- tion catalyst design. More specifically, the objectives of the work are: Introduction and Objectives 19
* To identify the location and morphology of nickel sulfide deposits on aged hydrodemetallation catalysts.
* To determine the deposition mechanism and the development of the deposition phenomena along the course of hydrodemetallation pro- cesses.
* To understand the interactions between the nickel sulfide deposits and the catalytic metals, and their implications to catalyst deactivation.
* To propose possible approaches to control the morphology of the deposits based on the results.
In the rest of the chapter, a short review of some subjects closely related with the present work will be discussed. Interested reader can refer to other comprehen- sive reviews on hydrodemetallation in general[85]. In chapter two, we will present a brief thermodynamic analysis of the system of a catalyst surface with metal sul- fide deposits, and attempt to understand the ultimate equilibrium deposition patterns on the hydrodemetallation catalysts. Chapter three describes the hydrodemetallation experiments by using model compound to simulate industrial hydrodemetallation pro- cess and to obtain spent and aged catalysts loaded with nickel deposits. In chapters four, five, and six, the results from electron microscopic studies for nickel deposition are presented. Chapter four shows the pattern and morphology of the nickel deposits on hydrodemetallation catalysts. Chapter five discusses the differentiation between two possible mechanisms for the deposition phenomena. Chapter six describes the metal distributions within the crystallites of deposits on the catalyst surfaces. Chap- ter seven is devoted to the mobility of the catalytic metal sulfides on the catalyst surface. In chapter eight, we attempt to depict a complete picture for the deposi- tion process and the implications to catalyst deactivation. Finally, chapters nine and Introduction and Objectives 20 Introduction and Objectives 20 ten summarize the main conclusions of the thesis and propose recommendations for future work, respectively. The main research topics are tabulated in Table 1.1.
1.3 Literature Review
1.3.1 Structure of Sulfided Co - Moly - A1203 Catalyst
CoO - MoOs3 / - Al2 03 is widely applied in industrial catalytic hydroprocessing for all kinds of petroleum feedstocks. Although it has been used for many years, and extensive efforts have been taken to explore the fundamental aspects of the catalyst, its atomic structure and catalytic mechanism are still in dispute. Many different models have been proposed to describe the structure of molybdenum and cobalt, and the mechanism of hydroprocessing reactions on the CoMo/A1 203 catalyst[23] [47] [51] [56] [121]. The metal oxide catalyst is usually presulfided to convert the oxides into sulfides. The sulfidation is necessary to prevent metal oxides from being reduced to metals, which are active for hydrogenolysis, thus could lead to rapid coke deactivation to the catalyst [5] [99] .
Figure 1-1 illustrates the complexity of a Co - Mo/ yA7 203 catalyst[91] [114], though it may still not be what a real industrial catalyst surface look like. For a typical sulfided Co - Mo/yA120 3 catalyst, there are many possible species on the catalyst surface. Typically, the main phase would be poorly or well crystallized
MoS 2, decorated with cobalt atoms on the edges of the MoS 2 layers. Co9S8 phase is also usually present. Other possible phases include CoSl+=, CoAl20 4, carbonaceous deposits, etc. IntroductionInruto and ObetieObjectives 21
Table 1.1: Research Objective
Specific Research Topics Characterization Chapters
. .
Hydrodemetallation Three
Deposition Patterns STEM, XRD Four, Six
Deposition Mechanisms STEM Five, Six, Two
Molybdenum Surface Segregation STEM, TEM, XPS Six, Two
Mobility of Catalytic Metals STEM, TEM Seven
Catalyst Deactivation STEM,TEM Seven, Eight, One
¥ _ . . , _ _ , _ , 22 Introduction and Ojectives I c ac
Co304 Co304
SULFIDING
CoL MoS2\-4.--- ba CoS a------MoS2
Figure 1-1: General Representation of CoO - MoO3/y - A 20 3 Catalyst Introduction and Objectives 23 Introduction and Objectives 23 The crystallites of molybdenum disulfide are very well dispersed on the alumina surface. High resolution electron microscope could identify the stacking of a very restricted number of layers of molybdenum disulfide. These MoS 2 structures are attached to alumina by either basal or edge plane. It is generally accepted that the active component for hydrodesulfurization and hydrodenitrogenation is molybdenum sulfide with cobalt as a promoter, though there is no general agreement on the structure and functionality of cobalt [17] [114] [82]. Voorhoeve et al. [119] [120] proposed an intercalation model from a solid state point of view. The model proposes that the Co is intercalated into octahedral sites at the edges between the MoS 2 slab (i.e., between the adjacent sulfur layers). The contact synergy model by Delmon et al. [22] proposes that cobalt sulfide exists as a separate phase (Co9 Ss) from which spill-over of hydrogen to the MoS 2 phase can occur. Finally, Topsbe and co-workers [114] have proposed the existence of the so- called Co - Mo - S phase as the predominant active species in promoted catalysts. The Co is thought to be located in the same plane as that of the Mo atoms, possibly in interstitial or substitutional posit ions. In addition to the synergistic effect and the Co - Mo - S phase theories, it has been suggested that the presence of cobalt inhibits carbon deposition, thus deceases deactivation rate[49] [131]. Another explanation is that cobalt is needed for keeping the dispersion of molybdenum sulfide on the surface[49]. For hydrodemetallation reactions, it is still unclear what is the active component, or more specifically, the active phases on the catalysts. Under thermal hydroprocess- ing conditions at temperature above 430°C, noncatalytic demetallation takes place as a result of sulfur-metal coordination and the attack of the nitrogen-metal bonds by activated hydrogen [19]. Takeuchi et al. [112] proposed a mechanism for catalytic hydrodemetallation reaction including a model of the active surface site to account for the directionally oriented growth of the V3S4 phase. A porphyrin type molecule Introduction and Objectives 24
releases its vanadyl to the sulfur on the vanadium sulfide surface. The vanadyl is then deoxygenated with H2S and forms a new sulfide surface to continue the growth process. It has been showed that the hydrodemetallation occurred via a sequential mecha- nism involving initial hydrogenation of peripheral double bonds to activate the por- phyrin, followed by a hydrogenolysis step which fragments the molecule and remove the metal[l] [122]. This suggests that there are at least two kinds of active sites on the catalyst surface. Ware and Wei [123] used dopants with different acidities to manip- ulate the acidity of the catalyst surface, as a result, the ratio between hydrogenation and hydrogenolysis reactions changed.
1.3.2 Deposition Patterns
The characterization of metal deposition on hydrotreating catalyst received relatively little attention in the past. It was only in the eighties that researchers began to study the morphology and structure of metal deposits by using electron microscopy and other techniques. Silbernagel et al. [103] [104] used nuclear magnetic resonance (NMR) and elec- tron spin resonance (ESR) to trace the deposition of vanadium onto CoMo/A1 20 3 from heavy oil feeds at 350°C. At low loadings (up to 0.7wt% vanadium), a vanadyl VO2 + species dominated ESR spectral components suggested that the V0 2+ ion was associated with defect sites on the alumina support. At higher vanadium loadings a diamagnetic vanadium species was observed by NMR. The irregularity of the ab- sorption signal suggested that the vanadium was present in a number of physically different sites, so a surface species was suspected. The maximum loading of this diamagnetic species was 5-10wt%. At yet higher loadings vanadium was present as a sulfide, probably V2S3. Electron microscopic analysis suggested the sulfide was present as crystallites. Introduction and Objectives 25 Inrucio an Obetie 25 By using electron paramagnetic resonance (EPR) analyses, Ledoux et al. [50][51] detected three different vanadyl species on a catalyst aged with vanadium porphyrin at 450°C, one with four nitrogen atoms, a second with four sulfur atoms, and a third with four oxygen atoms. A quantitative distribution between the three was given as 20%, 20% and 60%, respectively. Since the catalyst used was presulfided, no oxygen atoms should be found on the active phase. Therefore, about sixty percent of the vanadium was deposited on the support. They concluded that the vanadium is statistically dispersed on the full surface of the catalyst, both support and active phase. It should be pointed out that the vanadium porphyrin was impregnated on the catalyst, not by demetallatioin reaction, which might have caused the statistical distribution of vanadium on the full substrate. Loos et al. [57][58]compared the X-ray absorption fine structure (XAFS) spectra of pure V2 03 and the pseudo V203 phase soaked on 7 - Al 203 support. The two spectra exhibit considerable differences. It was concluded that the vanadium sulfide reacted on or with the support. Takeuchi et al. [112] used transmission electron microscopy and X-ray diffraction to analyze vanadium sulfide deposits formed by the hydrodemetallation of heavy oils. The deposits, which were believed to reside within the pores of the catalyst, were identified as vanadium sulfide crystallites with sizes of about several hundred angstroms to about one thousand angstroms. Toulhoat et al. [116] used a scanning transmission electron microscopy (STEM) fitted with an X-ray analyzer, transmission electron microscopy (TEM), electron mi- croprobe (EMPA), and X-ray diffraction analyzer (XRD) to analyze catalyst aged with a heavy industrial feedstock, pentane deasphalted Boscan crude. The deposits were identified to be vanadium sulfide (V3S4) with the presence of nickel. Deposited crystallite diameters observed were 20nm to 40nm near the edge of the catalyst and 5nm to 10nm near the center. However, they found that the number of crystallites Introduction and Objectives 26
did not change significantly from the edge to the center of the catalyst. Smith & Wei[107] [108] [109] studied hydrodemetallation with model compounds of nickel and vanadyl porphyrins with clean oil at 280 - 350°C. The study was con-
ducted with a commercial CoMo/A1 2 03 catalyst, HDS16A. The aged catalysts were studied extensively with transmission electron microscopy. Other techniques, includ- ing scanning electron microscopy, X-ray diffraction analyzer, X-ray photo-electron spectroscopy (XPS), were also used in the study. Smith found that, for a given hy- drotreating catalyst aged at a given set of operating conditions, the number of nickel sulfide crystallites remained relatively constant while the size of these crystallites grew with nickel sulfide loading. The sizes of these crystallites grew from 10nm to 15nm while the metal loading was increased from 37wt% to about 100wt%. The corresponding nucleation sites was estimated as 5 x 10-7A-3. Smith also studied another catalyst sample with very low loading of molybdenum (0.24wt%) and cobalt (0.68wt%). The numbers of crystallites were estimated at around 5 x 10-9-3, which is about two order of magnitude smaller than that of the HDS16A catalyst. It was suggested that the nucleation numbers of nickel deposits on the aged catalysts were related to the loadings of the catalytic components, e.g. molybdenum, cobalt, or phosphorus. Limbach [53] characterized catalysts aged with vanadium porphyrin. By using an- alytical electron microscope, he found that the crystallite size of the deposits increased with local loadings on the catalyst particles. In summary, considerable progress has been realized in the past ten years, but many questions remain to be answered. Metal sulfides generally deposit on catalyst surface as crystallites, though the possibility of a surface layer is not excluded. It is not clear whether deposition is a physical or chemical process, or is dependent on the metal loadings and hydrodemetallation conditions. Introduction and Objectives 27 1.3.3 Catalyst Deactivation
The deactivation of hydrodemetallation catalyst is a very complex process. Although many factors are contributing to the catalyst deactivation, the accumulation of metal deposits on the catalyst is the most important phenomena causing the deactivation, mainly due to the fact that deactivation caused by metal deposits are not regenerat- able. In industrial reactors, a catalyst bed may accumulate nearly double its weight in feedstock contaminants[85]. Obviously, deposits of this magnitude must severely affect the catalyst's ability to function. Various authors have studied the deactivation of hydrodemetallation reaction by using vanadium poisonous compounds[112], nickel poisonous compounds[128], or both [113], and have observed a very rapid deactivation at low coverage(<1.5% metal) followed by a much monotonous deactivation, and eventually a sharp decrease of activity. Similar results have been reported by others for both pilot experiments and smaller scale experiment[34][70][73]. Figure 1-2 shows one of the earliest results reported by Henke[34]. The results represent a temperature history of a reactor in order to maintain a constant sulfur level in the product. In other words, it represents the history of the catalyst activity. Most reports attributed the initial deactivation to the build-up of a steady coke loading on the catalysts[74][102]. However, Weitkamp et al [128] and Tamm et al[113] attribute the first stage to monomolecularlayer of nickel species depositedon the ac- tive Co-Mo-S sites and the second stage to the slow buildup of metal deposits layer by layer on the top of an initial monolayer laid down during the rapid deactivation period. They suggested that the catalyst activity is from the deposited species after the first monolayer deposits. Clearly, this does not explain the fact that hydrodemet- allation catalyst keeps a high activity far after the monolayer deposition, even though the activity of metal deposits is only about one third or less of that of the promoted catalyst[16][88][112]. There is generally no controversy on the third stage of the de- Introduction and Objectives 28 Introduction and Objectives 28
104
86 0
H 68
'.3l 50
32 2 4 6 8 10 12 14 Catalyst Age: Month
Figure 1-2: Deactivation of Hydrotreating Catalyst in Pilot-Plant Experiments[34] Introduction and Objectives 29
activation. It is generally agreed that the sharp activity declining was caused by the eventual pore plugging with the build-up of metal deposits. Ledoux et al[50] proposed that only a very small amount of vanadium was needed to poison the most active sites of Co or Ni promoted molybdenum catalyst by pref- erentially choose the octahedral cobalt sites and thus destroy the apparent synergy between Co or Ni and Mo.
1.3.4 Migration of Metals on Catalyst Surface
Migration of metals on catalyst surfaces is well studied as a sintering phenomena. Many investigators have studied the migration of nickel on different supports[3] [24] [46] [92]. Generally, temperature and the gaseous environment are two important factors to determine how the metal behaves.
Bogdanor and Rase [9] studied a NiMo/A1 2 03 hydrotreating catalyst aged com- mercially by a blend of heavy coke and virgin gas oil, without excessive metals. They found that the active components on the catalyst, nickel and molybdenum, were both mobile at reaction, regeneration, and sulfidation stages. It is expected that nickel de- posits would behave similarly under similar conditions. Pazos et al[74] have speculated that the deposited metals might migrate to the free alumina support to explain the maintenance of catalyst activity. Additional evidence is the findings of Fleisch et al. [25]. By using X-ray photo- electron spectroscopy, they found that the ratio of Mo/Al changes with the increase in metal deposits. They speculated that molybdenum may migrate to the top of the contaminated layers and remain exposed to reactants. Prasada et al[79] studied by X-ray photoelectron spectroscopy the surface enrich- ment of molybdenum on a multicomponent molybdate catalyst of the composition
50% Ni 3CosFe 3 BiPKO.lMo. 12 52 5 - 50%SiO 2 after being used in ammoxidation of propylene. The molybdenum signal increased by about 10%, while nickel and cobalt Introduction and Objectives 30
signals decreased by 20% and 10%, respectively. The effect of temperature on the deposit morphology and deposit structure on hy- drotreating catalysts has not been reported in the available literature. However, it is expected that the deposition patterns on catalysts would be affected by hydrodemet- allation temperature. Similar to sintering phenomena on catalysts, these deposits are also expected to migrate on the catalyst surface under certain conditions. Two distinct mechanisms for the growth of metal crystallites on supports have been proposed. A model based on particle migration and coalescence was published by Ruchenstein and Pulvermacher[94], while a model base on the transfer of metal atoms individually from one particle to another (interparticle transport) was proposed by Flynn and Wanke[26][27]. Hughes [36]summerizes that sintering has the following pattern. For very small particles (<20nm) growth occurs predominantly by particle migration. For larger particles, growth occurs by atom migration on the surface. Chapter 2
Thermodynamic Considerations
When one tries to rise above Nature one is liable to fall below it. -Sherlock Holmes, The Adventure of the Creeping Man Sir Arthur Conan Doyle
2.1 Introduction
In addressing the metal deposition on hydrodemetallation catalysts, we seek an atom- isitic understanding of the nickel distributions on the catalyst surface, and its inter- action with the catalytic metals originally on the catalysts. Among the questions to be considered are the following: What is the thermodynamic equilibrium state of the components on the catalyst surface? What determines the morphology of the crystallites of deposits? Are the deposits on the aged catalysts approaching thermodynamic equilibrium
31 Thermodynamic Considerations 32 state? We will attempt to approach these questions from several different perspectives in the coming chapters. In this chapter, we will just present some thermodynamic facts concerning the components on hydrodemetallation catalyst surface.
2.2 Phase Diagrams
As one can imagine, it is naturally difficult to construct a complete diagram for a ternary system in a two dimensional paper. On the hydrodemetallation catalyst, we have nickel deposition, cobalt and molybdenum, in either sulfided or unsulfided forms, excluding the effect of the existence of the substrate, and gaseous phase. Although the bulk thermodynamics probably inapplicable to catalyst surfaces and to supported catalysts, such data can still be employed in considering what might be the gross state of the catalyst or in determining a proper concentration of hydrogen sulfide in hydrogen to convert a catalyst to a desired state. In the following, we will present a few phase diagrams for the relevant systems. Although one can easily locate phase diagrams for two component systems in well organized literatures[2][61], three component phase diagrams involving solid phases are very scarce. Fortunately, Co-Ni-S system has been studied by some mineralogist[40][45]. One of the diagrams is shown in Figure 2-1. We tried in vain to locate a phase diagram for the system of Ni-Mo-S system. Figure 2-1 shows that at sulfur level below about 0.6, the corresponding cobalt sulfide and nickel sulfide forms almost complete solid solutions. It should be pointed out that the phase diagram is for 1273K, which is far above the hydrodemetallation temperature. Thermodynamic Considerations 33
S(l)
0.2 0.4 0.6 0.8 Co Ni Co mole%
Figure 2-1: Phase Diagram of Ni-Co-S at 1273K[40] .Thermodynamic Considerations 34 Thermodynamic Considerations 34 2.2.1 Systems with Hydrogen and Hydrogen Sulfide
For the present hydrodemetallation catalyst system, only the hydrogen reduction equilibria of the sulfides need to be considered inasmuch as these are the final equilib- rium states. Accordingly, the hydrogen reduction equilibria for the pertinent sulfides are shown in Figure 2-2, 2-3, 2-4[63]. In the hydrodemetallation temperature range of 600-800K, the metal oxide cata- lyst and nickel deposits are expected to be readily converted to sulfides, even with a small fraction of one percent of hydrogen sulfide. Molybdenum and cobalt should be
in the form of MoS 2 and Co9Ss as being reported in literatures.
2.3 Surface Segregation
The surface composition of alloys used in catalysis is in general different from the composition of the bulk, due to the difference in surface tensions between the two components. The problem of surface enrichment is of particular interest in the case of highly dispersed binary catalysts, composed of microclusters of metals, metal oxides, or metal sulfides in our system on carriers. There are several possibilities for the mi- crostructure of such systems. For a system with two constituents, When the two constituents are immiscible, separate microclusters of A and B on the carrier may be formed (Figure 2-5a). The other limiting case involves constituents of complete miscibility, when microcrystals of single phase solid solution are expected (Figure 2- 5b). There are then two possible microstructures with one component segregated to the surface: enrichment of one component in the surface layer with a nearly homo- geneous alloy at the center of the microcluster (Figure 2-5c), and separation of the crystal into two concentric phases of different composition, one on the inside and one on the outside(Figure 2-5d)[32]. Thermodynamic Considerations 35
3
2
1
0 I
-1
-2
-3
-4
-5 6 8 10 12 '14 16 18 20
1/T,K x104
Figure 2-2: Hydrogen Reduction of Nickel Sulfides[63] ThermodynamicThemoynmi Considerationsr 336
3
2
1
0
W~ -1
-2
-3
-4
-5 6 8 10 12 14 16 18 20
1/T,K x104
Figure 2-3: Hydrogen Reduction of Cobalt Sulfides[63] Thermodynamic Considerations 37 Thermodynamic Considerations 37
6
MoS3 4
2
bO 0 MoS 2
-2
-4
Mo
-6
-8 6 8 10 12 14 16 18 20
1/T, K x 104
Figure 2-4: Hydrogen Reduction of Molybdenum Sulfides[63] Thermodynamic Considerations 38
A (1) Ir ·" "·l·r" "l·lll· a
b
C
d
Figure 2-5: Possible Microstructures of a Highly Dispersed Alloy in a Substrate Thermodynamic Considerations 39 Throyai Considerations 39 Over 100 years ago, Gibbs[31] developed a comprehensive thermodynamic for- mulism for interface. The phenomena of surface segregation can be described in terms of that formulism by the use of the so-called Gibbs Adsorption Equation, which may be written for the case of an A-B binary system as:
dy = -SdT - rAdALA - FBdB (2.1)
where y is the surface energy, S' is the specific surface excess entropy, rA and rB
are the surface excess concentrations, and HA and HB are the chemical potentials of components A and B in the system, respectively. Applying regular solution model to Equation 2.1, one can obtain the following general result:
xurf ace XbulkA ep aGG (2.2) Xgur face XBUlkex RT(
where XAUtfacc and XuB face are the respective fractions of components A and B in the bulk phase, while Xgudk and XBj"k are the equilibrium fractions in the surface phase, and AG is the molar free energy of segregation. For solid systems, the free-energy, AG, is mainly composed of contributions from two terms: surface energy difference and strain release resulted from segregation, with the surface energy term as the dominant one [97] [101][129]. In a quasichemical approach for ideal solution where free energy AG can be ex- pressed in terms of the difference in bond energy between an A atom and a B atom AE as:
AG= AEi (2.3)
where i runs over all broken bonds of the surface atoms. Therefore, we would expect a compound with a cleavage plane consisting of very Thermodynamic Considerations 40 weak bondings should accordingly have low surface energy when the solid is cleaved from that particular plane. Chapter 3
Hydrodemetallation Experiments
The practical applicationof what I have said is very close to the problem which I am investigating. -Sherlock Holmes, The Adventure of the Creeping Man Sir Arthur Conan Doyle
3.1 Chapter summary
Hydrodemetallation experimental procedures and equipment were detailed in this chapter. The properties and specifications of the catalysts, model compounds, and other materials used in the experiments are presented. The planning of the experi- ments are also discussed. In addition to catalyst sample preparation, the characteri- zation techniques used in this study, including STEM, TEM, XPS, XRD, are briefly described. Characterization by scanning transmission electron microscopy (STEM) and high resolution transmission electron microscopy (HRTEM) showed that the de-
41 Hydrodemetallation Experiments 42
position of nickel sulfide on the catalyst surface enhanced the mobility of the catalytic components on the surface. The increarse of mobility was caused by the lowering melting point of Co9 Ss and MoS 2. The effect was discussed with conceptual mobility phase diagrams.
3.2 Introduction
A significant part of the metal compounds in petroleum comprises of poorly char- acterized organometallic molecules. In order to reduce the obscuring occurrence of competing catalytic and thermally induced reactions encountered with petroleum and residual feedstocks and other uncertainties, model compounds are usually used to con- duct kinetic and other laboratory studies. As a large part of metallic constituents in crude oil, petroporphyrins have been regarded as a suitable model compounds. Most of the reported hydrodemetallation studies have been performed with synthetic metal porphyrins[1][10] [15][39] [107][122][124] [128]. Compared with industrially aged catalysts, the laboratory aged catalysts are comparatively free of carboceneous de- posits. The metal loadings and aged conditions can also be easily controlled in order to obtain a clearer picture of the whole deposition phenomena.
3.3 Equipment
A one-liter batch autoclave reactor (Autoclave Engineers, Model AFP 1005) was used for the hydrodemetallation studies. The reactor system has been described previously by Hung [39] and Smith [106]. A schematic of the system is shown in Figure 3-1. The details of the autoclave are shown in Figure 3-2. A dual heating/cooling system in the autoclave allows a rapid isothermality of the reactor after the addition of feed at the initiation of each experiment. Modification to the reactor includes an addition of Hydrodemetallation.Experiments 43 HydrodemetallationExperiments 43~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ a sintered stainless steel basket to hold the catalysts inside the reactor. The nominal pore size of the basket is about 7m. Since the sizes of the porphyrin molecules are only about 1.0-1.2 nanometers [100], the basket does not block the access of the porphyrin molecules to the catalysts.
3.4 Model Compounds
Metal compounds in a crude oil are usually classified into two groups: porphyrinic and non-porphyrinic metal compounds. The non-porphyrinic part of the metal compounds are not yet to be well characterized, though the chemical information about the porphyrinic compounds are generally available. The latter usually accounts for about 10-50% of the metals found in crude oils [110]. For these reasons, metalporphyrins are regarded as suitable model compounds for studies on hydrodemetallation reaction. In this research, nickel etio-porphyrin, provided by Midcentury, (Posen, IL 60469), was used as the model compound for all the hydrodemetallation reaction. The molecular structures of the porphyrin is shown in Figure 3-3. Its solubility in squalane are about 20ppm at room temperature. At 588K, the solubility is unknown but higher than 300ppm [106].
3.5 Solvent
Squalane (2,6,10,15,19,23-hexamethyltetracosane) is used as the liquid carrier for metal porphyrins in the hydrodemetallation experiments. Squalane was supplied by Sigma Chemical Co., (St. Louis, MO). It consists of 97.4% iso-paraffins, a small amount of naphthenes and aromatics. It is free of sulfur, nitrogen, or metal com- pounds. It is in liquid state at room temperature, and has a relative high boiling point (>673K), so that the vapor pressure is very small at reaction condition. Some Hydrodernetallation Experiments 44 HydrodemetallatinExperiments 4
L
Autoclave
I - I I - - - -
Figure 3-1: Schematic of the Hydrodemetallation Equipment HydrodemetallationHyrdmtalto Exeiens4Experimentsr 45
Co
Figure 3-2: Schematic of the l-litre Autoclave Reactor Hlydrodemetallation Experiments 46 Hydrodemetallation Experiments 46
- C --
Nickel Etio-Porphyrin
Figure 3-3: Molecular Structure of Nickel Etio-Porphyrin Hydrodemetallation Experiments 47 Hyrd m talto Ex ei------e t ------47 of the important properties of squalane are listed in Table 3.1.
3.6 Catalysts
Catalysts used for this study are Co- Mo/7 - A120 3 provided by American Cyanamid Company. As supplied, the cobalt molybdate catalysts consists of a mixture of metal- lic oxides on an alumina support. These catalysts are activated by pretreatment at
atmospheric pressure with a mixture of 10 mol% H2S, 90 mol% H2. This pretreat-
ment converts the metallic oxides to metallic sulfides (MoS 2, CosS). The procedure used in this research was adapted from that recommended by American Cyanamid Company. Most of the previous work conducted in the MIT hydrodemetallation research
group was conducted on a commercial CoMo/A1 203 catalyst American Cyanamid Aero HDS16A[1][39][107][122].In the present work, this catalyst is used as the base case study. Its chemical and physical properties are listed in Table 3.2. Meanwhile, another catalyst was prepared by American Cyanamid Company specifically for this work. The chemical compositions of both catalysts are listed in Table 3.21. Both of the catalyst have the same element ratios. The molar ratios of molybdenum to cobalt on both catalysts are about 1.1.
3.7 Hydrodemetallation
The typical operating procedure was to load about 0.1 grams of catalyst of the size of 80/tm into the sintered stainless steel basket, and load charge the reactor with about 300grams of squalane. The catalyst was dried at 383K overnight before being loaded. The catalysts were presulfided in-situ by a 10 mol% hydrogen sulfide/hydrogen
1Thank Luhong for conducting the BET surface area measurement. Hydro demet allation Exp eriments 48 Hydrodemetallation Experiments 48
Table 3.1: Properties of Squalane
Supplier Sigma Chemical Co.
Lot Number 116F-0221
Chemical Formula 2,6,10,15,19,23- hexamethyltetracoane
Molecular Weight 422.8
Elemental
Sulfur < lppm wt Nitrogen lppm wt Nickel 0.25ppm wt Vanadium < 0.05ppm wt Hydrogen < 15.29 wt%
Density at 273K 0.80610
Viscosity at 313K 19.26cS
Viscosity at 373K 4.149cS
P/N/A Distribution (wt%)
Paraffins 97.40 Mono Naphthenes 0.00 Poly Naphthenes 1.75 Aromatics 0.85
Simulated Distillation by GC Analysis
IBP 687K 50% 704K 90% 708K 95% 708K FBP 716K Hydro demet allation Exp eriments 49 Hydrodemetallation Experiments 49
Table 3.2: Compositions of Catalysts
Mo wt% Co wt% P wt% }Surface Area (m 2 /g)
HDS16A 8.13 4.48 3.0 176
SN6931 5.09 2.87 2.21 171
', ~ ~ ~ ~. . . . L -- Hydrodemetallation Experiments 50 Hydrodemetallation Experiments 50
Table 3.3: Properties of Catalyst HDS16A
Supplier American Cyanamid Co.
Lot Number MTG-S-0573
Chemical Properties:
CoO 5.7wt% (dry basis) MoO 3s 12.2wt% Na 2O 0.03wt% Fe 0.04wt% Ni 0.O9wt% Si 0.15wt% P 3.00wt% Al 20 3 base
Physical Properties:
Average diameter 0.152cm Average length 0.432cm Pore volume 0.43ml/g Surface area 176m2 /g Particle density 1.49g/cm3 Median pore diameter 80.4 A Hydrodernetallation Experiments 51 Hrdmtalto Exeiet 51 gas mixture (Matheson Gas Products). After pressure testing, the reactor was purged for about 0.5 hours under a flow of helium(99.995% purity, Matheson Gas Products).
Sulfiding was achieved with a mixture of 10 mol% H2 S/H2 (Matheson Gas Products) flowing at a rate of about 200ml/min, according to the standard temperature program. The temperature was held at 448K for six hours, before being raised to 588K at a rate of 60K/hour, then maintained at the temperature for one hour. Operating conditions for the hydrodemetallation experiments ranged from 588K to 623K at hydrogen pressure of 4.8mPa. The partial pressure of H2S was maintained at about 14kPa (0.3vol.%), though it was not precisely controlled. The gas samples were routinely analyzed for hydrogen sulfide concentrations using gas detector tubes (Kitagawa, Japan, H2S 1-150ppm). Prior to each hydrodemetallation experiment run, the reactor was pressurized with
a mixture of 10 mol% H2S/H 2 and then hydrogen to pressures which would achieve the desired hydrogen sulfide partial pressure and total hydrogen pressure. Then, a slurry of nickel porphyrin in about 100ml squalane was added to the preheater. The preheater was then purged under the flow of helium before being heated to the same temperature as that of the reactor. Hydrogen was then introduced to the preheater to a pressure a little higher than the reactor pressure. The slurry is then injected to the reactor by open the valve between the reactor and preheater. The procedure was repeated twice with about 100ml squalane to ensure that no undissolved nickel porphyrin was left in the preheater. The flushing was later found essential because of the low solubility of nickel etio-porphyrin in the solvent. Even with the repeated flushing, it was found that some undissolved porphyrin remained in the bottom of the preheater. As a consequence, material balances for nickel were not obtained for most of the runs. The nickel loadings in the subsequent chapters are all referring to the nickel loadings actually obtained through atomic adsorption analyses. Some of the hydrodemetallation runs were performed in a second reactor system. Hydrodemetallation Experiments 52 yt E 52
The system has a two-liter autoclave reactor. It also allows a constant flow of 10 mol% H2S/H 2 mixture, thus a better control of the hydrogen sulfide concentration in the system. Catalyst HDS16A was used to study the development of deposits on the catalyst surface. Hydrodemetallation experiments was also conducted without hydrogen sul- fide in the system and with unsulfided catalysts. The purpose was to study the form of different deposits on the catalyst surface. Table 3.4 is a summary of the hydrodemetallation runs for which the aged catalyst samples were characterized by various techniques. Note that the metal loading are all at about 15-20%. The nickel loading is defined as the amount of nickel on fresh catalyst bases. The relatively lower metal loadings was chosen to avoid the domination of one compound over the others on the catalyst surface. It would be difficult to study the interaction if the nickel loading is either much higher or much lower than the cobalt and molybdenum loadings on the original catalysts.
3.8 Characterization
To prevent the aged catalyst samples from air exposure, both Smith [106] and Limbach [53] transfered the aged samples to a gloves-box filled with argon under the cover of oil before the preparation of characterization samples. The oxidation of the metal sulfide at ambient conditions is a relatively slow process [25] [118]. X-ray photoelectron spectroscopy showed that a minimal sulfur oxidation is observed after exposure to air for a week [25]. Nevertheless, all aged catalyst samples with the stainless steel baskets were transfered to a gloves-box filled with argon, and repeatedly washed with xylene and acetone before being dried in a self-sealing quartz crucible (Fisher Scientific). The catalysts are then ready for preparing any samples for characterization. The major characterization tool was electron microscopy, including high resolu- Hydro demet allation Exp eriments 53 Hyrdmtalto Experiments 5
Table 3.4: Summary of Hydrodemetallation Runs
HDM Conditions Aging Times Ni Loadings run Catalysts
T PH| PH2S (Hrs.) (wt.%) __. (K) (MPa) (KPa)
1 HDS16 623 4.8 14 650 23%
2 HDS16 623 4.8 0 670 22.6%
3 SN6931 588 4.8 14 380 22.1%
4 HDS16 648 4.8 14 200 0%
.. _ - . Hydrodemetallation Experiments 54
tion transmission electron microscopy (HRTEM), and scanning transmission electron microscopy (STEM). The high resolution transmission electron microscopy allows us directly observed the structure of the deposits, while the scanning transmission elec- tron microscopy offers a unique approach for measuring individual small crystallites which may be catalytically active as opposed to the averaging method employed in spectroscopic techniques. During electron microscopic analysis, contamination of the surface of the speci- men can be produced by the electrons polymerising hydrocarbons adsorbed on the surface from the residual gases in the vacuum. Contamination can also appear if there are residual oils on the specimen as in catalysts for hydrodemetallation in our system. Therefore, the repeated washing of the samples and careful handling with the specimen during microtome are essential to avoid the contamination during electron microscope analysis. The sample preparation for electron microscopes was completed by embedding catalyst sample in resin, and ultramicrotoming to get specimen with the thickness of 60 to 80nm slices. The embedding medium was an ultra-low viscosity resin provided by Ladd Research Industries, Inc. The composition of the resin is listed in Table 3.5. The detailed procedure for preparing the specimen is as follows:
1. A very small amount of aged catalyst particles are dispersed in a plastic em- bedding capsule. Any chucky clusters would be carefully blown away with a dust chaser. The particles should be in a very well dispersed layer on the bot- tom. The amount of catalyst particle should be as small as possible. Excessive amount of particles would cause difficulty to get complete specimens during microtoming.
2. Slowly pour the well-mixed resin into the capsules, and let the capsules sit overnight in a desicator for better infiltration. The resin was then cured in an Hydrodemetallation Experiments 55
Table 3.5: Ladd Ultra-lowViscosity Embedding Medium
Weight , Materials Weight Percentage
2.5g 4-vinylcyclohexene dioxide (VCD) 31.85%
5.25g n-Hexenyl succinic anhydride (HXSA) 66.88%
Og Diglycidyl ether of polypropylene glycol (DER-736) 0%
0.1g Dimethylaminoethanol (DMAE) 1.27%
I ~~ ~ ~ ~ ~ Hydrodemetallation Experiments 56 HydrodemetallationExperiments 56~~~~_ oven by slowly heating up to 333K and maintaining the temperature for 3 to 5 days. Some samples were cured at 333K for 10 hours, followed by 24 hours curing at 353K. Both of the curing procedures were found adequate for getting good block quality for microtoming.
3. The embedded samples were trimmed with a self-prepared glass knife into a trapezoidal shape on an LKB Ultratome III machine, with a face containing the specimen exposed to the knife for slicing. The face should be as small as possible to avoid unnecessary knife damage. Diamond knife was then used to cut the sample to get thin specimens with the thickness of about 60 to 80 nanometers. The microtome was conducted by following the procedure recommended by Jones[42]. The article by Rice & Treacy[90] also contains useful information on ultramicrotomy. Figure 3-4 schematically shows the slicing of samples during microtomy.
4. The specimen film is then supported on a copper grid and then coated with carbon for enhancing electron conductivity. The samples for TEM and STEM are virtually the same, though a thicker carbon coating was needed for STEM to allow the X-ray analysis.
The sample preparation for XPS and XRD are comparatively much simpler. The XPS sample was prepared by pressing the aged catalyst particles into an indium foil. The XRD sample is mounted in cement. Total metal contents on the catalysts were analyzed at GalbraithLaboratory, Inc. by atomic absorption spectrophotometry (AAS). Next, we will briefly discuss each of the techniques used in the studies, and show their uniqueness in characterizing catalyst samples. HydrodemetallationHyrdmtalto ExeiensExperiments 57
_ _
Catalyst Particles
A. Side View
Slices
B. Top View
Figure 3-4: Schematic Diagrams of Ultramicrotomy Hydrodemetallation Experiments 58 Hydroemetllaio Exermet 58 3.8.1 High Resolution Transmission Electron Microscope (HRTEM)
High Resolution Transmission Electron Microscopy is the only technique that make possible the direct description of the microstructure of solids in real space. In a transmission electron microscope, a thin specimen is irradiated with an elec- tron beam of uniform current density. Electrons are emitted in the electron gun by thermionic emission or by field emission. A two stage condenser-lens system permits variation of the illumination aperture and the area of the specimen illuminated. The electron-intensity distribution behind the specimen is imaged with a three or four stage lens system, onto a fluorescent screen. The image can be recorded by direct exposure of a photographic emulsion inside the vacuum. Figure 3-5 illustrates the interaction of electron with a specimen. The electrons interact strongly with atoms by elastic and inelastic scattering. The specimen must therefore be very thin, typically of the order of a few tens up to a few hundred nanometers, depending on the density and elemental composition of the object and the resolution required. TEM can provide high resolution because elastic scattering is an interaction pro- cess that is highly localized to the region occupied by the screened Coulomb potential of an atomic nucleus[89].
3.8.2 Scanning Transmission Electron Microscope (STEM)
As the name scanning transmission implies, an electron probe, formed by an objective lens incident on a thin specimen is scanned across it, and either the directly trans- mitted (bright field) or scattered (dark field) electrons are collected by an annular detector, whose output modulates a display scanned in synchronism with the signal in the scanning coils in the instrument. An energy analyzer is used to give elemen- Hydrodemetallation Experiments 59
Incident electron beam
Backscattered primary electrons X-ray
Auger electrons Light Secondary electrons
Thin Specimen
tered electrons Elastical
Unscattered electrons
Figure 3-5: Signals Created by the Interaction of High Energy Electrons with the Specimen Hydrodemetallation Experiments 60 Hydodeetlaio Exeiet 60 tal analysis from specific energy losses. For a dedicated STEM, the scanning spots can be down to lnm in diameter, obtained by demagnifying a field emitter source of electrons. STEM is most attractive for a number of reasons. Firstly, it can be used to select very small (1-2nm) individual crystals, and give their diffraction patterns and X-ray emmision spectra, or characteristic X-ray lines. There is a simple relationship between X-ray frequency v (or energy) and atomic number Z given by
v = 0.248(Z - 1)2 x 1016 (3.1)
Hence measurement of the energy of one of the X-ray emissions from an element allows that element to be identified. Since the image of a thin specimen is always in view on the fluorescent screen, the area chosen for analysis can be located very accurately. An even greater advantage of thin specimens is that the resolution of the transmission image, which is about 10 -20 times greater that that of a scanning electron microscope image of a bulk specimen, allows the analysis to be correlated with fine details of ultrastructure[95][98]. The spatial resolution for analysis in a STEM system may well be limited by the stability of the specimen. Very small particles may move by more than the beam size during the time required to accumulate counts for analysis, and so for this application it is most important that the stage of the instrument be as free from drift as possible[30].
3.8.3 X-ray Photoelectron Spectroscopy (XPS)
X-ray photoelectron spectroscopy uses monochromatic X-rays such as Mg Ka (1254.6eV) or Al K, (1486.6eV) as an energy source to irradiate the sample. Core electrons in the target with binding energy less than that of the incident photon can be ejected HydrodemetallationHyrdmtalto ExeietExperiments 61 and the energy spectrum of these photoelectrons are analyzed in a multiplier. Since the mean free paths of these low energy electrons are relatively small, typically below 5nm, XPS signals are very surface sensitive. Since the energies of inner electrons are characteristic of the atom concerned, the identification of the atomic species present at the surface region of a solid may be carried out in a straightforward manner by X-ray photoelectron spectroscopy. One can identify the existence of specific elements by comparing the excitation lines with standard spectra. One important application of the XPS is its ability for depth profiling, which can be accomplished by controlled erosion of the surface by ion sputtering. With alternating sputtering and XPS analysis, it can provide a concentration profile within the outmost layer of a material.
3.8.4 X-ray Diffraction Analyzer (XRD)
X-ray diffraction was conducted for the aged catalyst samples to identify the crystal structure of the nickel deposits on the catalyst surfaces.
3.8.5 Surface Area Measurement(BET)
The surface area of the catalysts were measured by nitrogen desorption with the standard one-point B.E.T. method. Chapter 4
Nickel Deposition on Co- Mo Catalyst
The world is full of obvious things which nobody by any chance ever observes. - Sherlock Holmes, The Hound of Baskervilles Sir Arthur Conan Doyle
4.1 Chapter Summary
Nickel deposition from nickel-etio porphyrin on sulfided and unsulfided CoMo/yA1 20 3 hydrodemetallation catalysts were investigated. The bare and aged catalysts were characterized by high resolution electron microscope, scanning transmission electron microscopy and X-ray diffraction analyzer. Nickel deposits are found in crystallite
1This chapter is a revision of reference[127] J. Wei & X. Zhao, Chem. Eng. Sci. 47, 2721, 1992.
62 Nickel Deposition on Co - Mo Catalyst 63 Nicke C DepoitiononMoCtlyt6 - form. The average sizes of the crystallites are about 10 to 15 nanometers at the nickel loading of about 20%. Both element mapping and XEDS microanalysis from scanning transmission electron microscope showed that the elemental distribution of deposited nickel were correlated with elemental distributions of cobalt, but not with elemental distributions of molybdenum.
4.2 Introduction
Although electron microscopic investigation of hydrotreating catalysts has been a subject of many researches[20] [21], studies on aged catalysts are still very scarce[108]
[112] [116]. Metal deposition patterns on hydrodemetallation catalyst has been a subject of wide speculations. Uniform layer deposition was the most widely assumption un- til Toulhoat et al.[116] and Smith & Wei[108] directly showed by using microscopic techniques that nickel deposits are in segregated large crystal`te forms. Smith & Wei suggested that the number density of nickel deposit crystallites might be related with the loadings of catalytic metals on the catalyst surface. However, they did not speculate with which component that nickel might be associated.
4.3 Electron Microscopy Results
In the microscopic studies reported in this and subsequent chapters, care was taken to ensure that the area and particles studied and analyzed are representative of the catalyst samples. Samples are usually examined at low magnification for any inho- mogeneities. The characterization was conducted on two electron microscopes. One was a dedicated scanning transmission electron microscopy (VG HB5, Vacuum Generator) Nickel Deposition on Co - Mo Catalyst 64 64 Nickel Deposition on Co - Mo Catalyst equipped with Link energy dispersive X-ray analyzer. X-ray mapping and chemical microanalysis were obtained at 100keV, with a nominal probe size of about 2nm. The other was a high resolution transmission electron microscope (Akashi Topcon EM002B, Akashi Beam Technology Corporation) with a point to point resolution of 0.18nm(200keV).
4.3.1 Bare Catalyst
Figure 4-1 is a micrograph taken on STEM of a HDS16A catalyst sample. The catalyst sample was sulfided but was not subjected to hydrodemetallation. We could see from the picture the existence of the alumina platelets of the catalyst substrate. The sizes of these alumina platelets are estimated to be about 100 nanometers in length and 5 nanometers in width. No structures of the molybdenum or cobalt sulfides could be observed due to the resolution of the instrument. An X-ray elemental mapping of the catalyst did not reveal much details either (Figure 4-2). However, The mapping results did tell us that both cobalt and molybdenum are very well dispersed on the substrate. Figure 4-3 is a micrograph from high resolution transmission electron microscope for the same catalyst sample. Consistent with literature reports[18][20][80][96], we could see many randomly oriented molybdenum disulfide slabs. The lengthes of the slabs usually exceeds 10 nanometers. The slabs have usually four to five layers, but it would be difficult to observe the slab of one or two layers. No cobalt sulfide was observed on the surface. According to literature[20][21], the difficulty of observing
Co9 S8 may be associate with the low intensity of the main reflections of CoSs crys- tals. Comparatively, the existence of a strong (002) reflection for MoS 2 made it relatively easier to be observed and to be distinguished from other crystals. Thus, the absence of electron microscopic evidence does not rule out the existence of tiny
Co9S8 crystals. Nickel DepositionNickel on Cooo - Mo CatalystDl 65
,,rI 6 n,"'N n+v" "'A', I#, ~'
Scale: 100nm
Catalyst: HDS16A Temperature:
Aging Time: Ohrs. H2 Pressure:
Figure 4-1: Electron Micrograph of Bare HDS16A Catalyst NickelNickel Deposition ononCo Co - MMoCatalyt6 Catalyst 66
STEM _ TEM Micrograph Molybdenum 8.13% I
Nickel 0% Cobalt 4.48%
ii Scale: 50nm
Catalyst: HDS16A Aging Time: 0 hrs.
H2 Pressure: Temperature:
Figure 4-2: Elemental Mapping of Bare HDS16A Catalyst Nickel Deposition n o - Mo Catalyst 67 Nicke DepoitiononC~ - MoCatalst 6
______
Scale: 5nm
Catalyst: HDS16A Temperature:
Aging Time: 0 hrs. H2 Pressure:
-~~, -.-....
Figure 4-3: High Resolution Image of Sulfided Bare HDS16A Catalyst Nickel Deposition on Co - Mo Catalyst 68 Nike Deoito onC oCaayt6 4.3.2 Sulfided Catalysts
Figure 4-4 is the micrographic image of a sulfided catalyst sample (SN6931) aged at 588K. One dominant feature is the presence of many dark spots, which XEDS confirm to be nickel sulfide crystallites. Figure 4-5 is a set of element mapping images of the central area of the sample in Figure 4-4, which gives us not only the sizes and numbers of the deposition crystallites, but also direct information of the locations of each element and their interrelations. Although many crystallites are as large as 30 to 40 nm, there are also many crystallites of about 0.5nm. The average sizes of the crystallites were very roughly estimated to be about 10 to 15 nanometers. One important finding is the association between cobalt and nickel on the mapping. Particularly, we can see a distinct nickel crystallite at lower left side of the map, and there is a corresponding cobalt site, but no detectable molybdenum was seen at that site. In addition, we can also see that the cobalt crystallite on the catalyst seems in discrete forms, but the nickel deposition crystallite come together and form large entities. It suggests that at high metal loading, the neighboring deposition crystallites could coa lesce. Figure 4-6 is a set elemental mapping results for an aged HDS16A catalyst with about 23% nickel loadings. The association between nickel and cobalt is apparent. The nickel and cobalt maps well correspond to the black spots on the TEM micro- graph. The large crystallite in the left side of the picture will be further characterized for elemental distribution in Chapter six. STEM has a fine probe beam size of about 2nm, while the sizes of many deposi- tion particles are about 10nm or even larger. Therefore, it was possible to place the beam on some distinct crystallites to analyze the compositions of individual crystal- lites. Unfortunately, we can not determine the absolute composition for each par- ticular analysis, since we do not know the amount of alumina and other elements in this particular area. Nevertheless, we can still get the relative compositions of each Nickel Deposition on Co - Mo Catalyst 69 Nickelon Co Deposition- Mo Catalyst 6
Scale: 50nm
Catalyst: SN6931 Temperature: 623K
Aging Time: 380hrs. H2 Pressure: 4.8 MPa
Figure 4-4: Electron Micrograph of Aged Sulfided SN6931 Catalyst Nickel Deposition on Co - Mo Catalyst 70 k i o n Col 7
______STEM ______TEM Micrograph Molybdenum I 5.09%
I
Nickel 22.1% Cobalt 2.87%
_ - -
Scale: 100nm
Catalyst: SN6931 Aging Time: 380hrs.
H2 Pressure: 4.SMPa Temperature: 588K
Figure 4-5: Elemental Mapping of Aged SN6931 Catalyst with Sulfur Nickel Deposition on Co - Mo Catalyst 71 71 Nickel Deposition on Co - Mo Catalyst
STEM
_ __ TEM Micrograph I Molybdenum 8.13%
|I Nickel 23% C obalt 4.48%
Scale: 100nm
Catalyst: HDS16A Aging Time: 650hrs.
H2 Pressure: 4.8MPa Temperature: 623K
Figure 4-6: Elemental Mapping of Aged HDS16A Catalyst with Sulfur Nickel Deposition on Co - Mo Catalyst 72 72 Nickel Deposition on Co - Mo Catalyst crystallites. Although most of the analyses were conducted by focusing the beam on distinct crystallites, some analyses were conducted by purposely positioning the beam on some featureless background. Figure 4-7 and 4-8 are plots of the normalized compositions of the deposited nickel and the two catalytic metals, cobalt and molyb- denum in triangular and cartesian plots. The line in the triangular plot represents the molybdenum/cobalt ratio on the original catalyst. It is clear that those sites with lower Co/Mo ratio have a lower concentration of nickel deposition, while those sites with higher Co/Mo ratio show a much higher nickel concentration, indicating that most of the nickel is associated with cobalt, but not with molybdenum. The error bar represents a typical intrinsic error of the EDS technique calculated from the counts when conducting analysis. The procedure worked out by Furdanowicz[28] was used for the calculation of multicomponent analysis. The trends are significant. Emphasis was placed on identifying qualitative trends rather than quantitative results which would be difficult to support because of the variations within the small samples. It should be noted that each sample on the X-ray microanalysis plot corresponds to one particular sites on the catalyst surfaces. The absolute composition of each particular element is unknown since the amount of other elements are not available. Therefore, the absolute amount of the elements are not comparable for the different analyzing sites. Nevertheless, we can still see a clear tendency of the association between Co and Ni. As we have pointed out, The aim is to develop physical insight and to recognize trends, rather than to explain every observation.
4.3.3 Unsulfided Catalysts
Figure 4-9 shows a set of elemental maps of unsulfided HDS16A aged at 623K. The image of nickel deposition shows a striking correspondence to that of the cobalt. Some of the interesting locations are marked on the mapping pictures. While there is no apparent correspondent molybdenum at location 3, the images of nickel and cobalt NickelNike DeoiinonCDeposition on Co - MooCaayt7 Catalyst 73
Ni
Co Mo Catalyst: HDS16A Temperature: 623K Hydrogen Pressure: 4.8MPa Nickel Loading: 23%
Figure 4-7: EDS Microanalysis of Aged HDS16A Catalyst Nickel Deposition on Co- Mo Catalyst 74
_ E_· _ - DuI- -r
SulfidedIHDS16 ik, 623KI 50 .i.picalE..... r B ITpical Prwrr Bo *1... 0o
40 ......
0 i .. . :. .I . . . . .
30 .,..... ;
......
20 -················ · · · · · · · · · · ......
...... ; . .
10 O ..... -...... 0. .: ...... 0 0 0 : I o 0: n -d -I i i 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Co/(Co+Mo)
Figure 4-8: EDS Microanalysis of Aged HDS16A Catalyst Nickel Deposition on Co - Mo Catalyst 75
are apparently similar. The image of molybdenum is more or less randomly dispersed on the whole surface. In addition, it was observed that the distribution of iron on the catalyst was very well corresponded to that of cobalt. The iron was probably from the sintered stainless steel basket holding the catalyst during the hydrodemetallation reaction. Similarly, microanalysis was also conducted on the unsulfided aged catalyst. The result is shown in Figure 4-10 and Figure 4-11. Apparently, the trend is quite similar with the one observed on sulfide catalyst.
4.4 X-ray Diffraction Results
The X-ray diffraction was conducted on a Rigaku diffractameter. Figure 4-12 shows the X-ray diffraction spectra for an aged HDS16A catalyst sample, which has about
23% of nickel loading. The spectra clearly indicate the existence of bulk phase Ni 7S6, although we were expecting Ni3 S2, as being obtained from. the work of Smith & Wei[108]. The difference is probably a result of the inaccurate control of the partial pressure of hydrogen sulfide for the hydrodemetallation experiments.
It is peculiar that there is no Ni 7 S6 in the reduction phase diagram we presented in chapter two. One literature[48] did indicate that the diffraction pattern of Ni 7S 6 was very similar to the earlier reported diffraction pattern of Ni 6S5 by Lundqvist[59]. Considering the data used for the phase diagram shown in Figure 2-2 was reported in
1954 [93], the Ni 6 S5 phase in the phase diagram could be simply a misidentification.
4.5 Discussions
There were a lot of speculations about the possible deposition patterns of nickel deposits on the catalyst surface, e.g. layer deposition, random deposition, crystallites. Nickel Deposition on Co - Mo Catalyst 76
Figure 4-9: Elemental Mapping of Aged HDS16A Catalyst without Sulfur Nickel Deposition n Co - Mo Catalyst 77~77 NickelDeposition on Co - Mo Catalyst
Ni
Co Mo Catalyst: HDS16A Temperature: 623K Hydrogen Pressure: 4.8MPa Nickel Loading: 22.6%
Figure 4-10: EDS Microanalysis of Aged HDS16A Catalyst without Sulfur Nickel Deposition on Co - Mo Catalyst 78
ItA 14 l f . , . . .,
:0 12 ...... I......
0
...... 10 .... I. I...... I...%...... I.. ... ; ...... o
R ...... o 8 0 :
I .I 0 6
...... 0... :rT__lCI. . an,I I 4 ...... I . .... Iunsuinaeai...... 5A,623K
0
2 .I...... 11......
0 ,o 0
i III !.t . i I _ 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Co/(Co+Mo)
Figure 4-11: EDS Microanalysis of Aged HDS16A Catalyst without Sulfur Nickel Deposition on Co - Mo Catalyst 79 ______
625
500
375
250
125
0 2 THETA r
Catalyst: HDS16A Temperature: 623K Aging Time: 650hrs. Pressure: 4.8 MPa
,~~~ , . i ·
Figure 4-12: X-ray Diffraction Spectra of Aged HDS16A Catalyst Nickel Deposition on Co - Mo Catalyst 80 Nickel Deposition on Co - Mo...CatalystI I 80II
Smith & Wei[108] suggested that the nucleation sites for nickel deposits on the catalyst surface might be related with the loadings of catalytic components Co, Mo, P, on the catalysts. Among all the possible deposition sites, cobalt is the less expected sites for nickel deposits, due to the fact that molybdenum sulfide is the active sites. We will address the mechanism for the deposition and the reason that nickel is associated with cobalt on the catalyst surface in the coming chapter. Here we want to emphasize that the observation itself is a very significant result in a sense that we, for the first time, know that nickel deposits are not randomly distributed are the catalyst surface. Instead, it forms identities with cobalt sulfide on the catalyst surface. The understanding of the structure of the deposits maybe well be the first step for the development of improved hydrodemetallation catalysts.
4.6 Conclusions
Catalytic hydrodemetallation of model compound nickel etio-porphyrin was con- ducted with presulfided Co - Mo/7/AI203 catalysts. The catalysts before and af- ter hydrodemetallation were characterized by using scanning transmission electron microscope, high resolution transmission electron microscope, and X-ray diffraction analyzer. On a separate experiment, an unsulfided catalyst aged under similar con- dition is also characterized by scanning transmission electron microscope.
1. On the sulfided catalyst before being subjected to hydrodemetallation, we could observe the layer structure of molybdenum sulfide. The crystals usually consists of about four to five layers of S-Mo-S, with the average lengths on the order of a few nanometers. No crystals of cobalt sulfide were observed.
2. Nickel deposits on hydrodemetallation catalyst in crystallite form. The sizes of these crystallites vary in a wide range. At the metal loading of about 20%, the Nickeli DDeposition I i on Co - MoMo CCatalyst l 81
largest ones observed are around 40nm, while the smallest ones detected are around 0.5nm. The average sizes were roughly estimated to be about 10 to 15 nanometers.
3. Elemental mapping and XEDS microanalysis show that nickel deposits are pref- erentially associated with cobalt for both sulfided and unsulfided systems. This is a very significant result for the following reasons. First of all, the results for the first time directly shows that the deposition is neither a uniform layer distribution as being modeled[43][55][86], nor randomly distributed on the cat- alyst surface, as has been suggested[125]. Secondly, since molybdenum sulfide is considered the active sites for all the hydrotreating reactions, including hy- drodemetallation, while cobalt sulfide is only a promoter on the catalyst. The fact we showed that nickel deposits are associated with cobalt instead of molyb- denum could have significant implications on the deactivation of hydrodemet- allation catalysts. This subject will be pursued further in the coming chapters. Chapter 5
Deposition Mechanism
We imagined what might have happened, acted upon the supposition, and find ourselves justified. - Sherlock Holmes, Silver Blaze Sir Arthur Conan Doyle
5.1 Chapter Summary
Two possible mechanisms for the associations of nickel and cobalt are proposed: ac- tivity of cobalt site, or migration after deposition. To differentiate the two deposition mechanisms, CoMo/yAl203 catalyst with 10% impregnated nickel was treated under hydrotreating conditions without nickel-porphyrins. It was observed that the nickel on the catalyst surface migrates towards cobalt sites, though the initial deposition was uniformly distributed on the surface without any preference to either cobalt or molybdenum. Previous activity studies on cobalt-only or molybdenum-only catalysts
82 Deposition Mechanism 83 support the theory that molybdenum sulfide, not cobalt sulfide, is acting as the main active component.
5.2 Introduction
The association of nickel with cobalt on the elemental mapping could have two differ- ent implications. Either the activity of cobalt sites or the migration of nickel deposits could have been caused the association. The thermodynamic analysis in chapter two favors the argument of migration of deposits and formation of solid solution. However it did not exclude the possibility of the activity of cobalt. We found a previous work by Hung [38]. Hung conducted kinetic studies on Co/ 7 A1 203 catalyst, Mo/yAl 2 0 3 catalyst and CoMo/-yAI20 3 catalysts for hydrodemetallation reaction, respectively. It was shown that the Mo-catalyst is much more active than Co-catalyst, although both of them are active for the hydrodemetallation reaction, as shown in Figure 5.2. Therefore, it is unlikely that cobalt solely is acting as the active entity for hydrodemet- allation on CoMo/ 7 A1 203 catalysts. A plausible explanation for the association of nickel with cobalt is that the nickel sulfide preferentially migrates to cobalt sites after being deposited on the catalyst surface. Considering that the Tammann temperature for nickel sulfide is only about 530K, which is much lower than the hydrodemetallation temperature of about 588-623K, nickel sulfide should be very mobile. The mobility of nickel sulfide on hydrotreating catalyst was reported in literature [9]. It should be pointed out that Hung's work [38] was conducted in unsulfided sys- tems. Deposition Mechanism 84
rMf law I I lll Ill I II I I I I I I I I I II~~~~~~~~~~~~~~~~
Mo 10% ¢ ) Co 0% 500 Mo 10.1% Co 3.63% 0 0 400 Mo 10.0% Co 5.5% K 300 - 0 Mo 10.0% Co 9.0%
~ m A A~ ~ 200 Mo 0.0%( ) Co 5.7%' I
100
0 I I I 0.0 0.2 0.4 0.6 0.8 1.0
Co/(Co+Mo) Molar Ratio
Figure 5-1: Activity of Catalysts with different Cobalt Contents[9] Deposition Mechanism 85 Deoito Mehais 85 5.3 Migration Experiment
To further confirm the migration of nickel deposits on the catalyst surface, the fol- lowing experiment was designed. Instead of hydrodemetallation reaction, nickel was impregnated on the CoMo catalyst. The catalyst with nickel was put back to the re- actor under the same condition with hydrodemetallation reaction condition, but with no nickel porphyrin added. The aim was to observe the development of the nickel on the surface. The impregnation procedure was the same as the way to making nickel catalyst, the procedures is as follows[38]:
1. The HDS16A CoO - MoO3/AI20 3 catalyst was crushed into 170-200 mesh size (d=0.081mm), and then placed in a tubular furnace at 713K to for 24 hours to remove water.
2. Nickel solution was prepared from nickel nitrate, Fisher Scientific Co., Fair
Lawns, N.J.; Ni(NOs3) 2 * 6H20; F. W. = 290.81; NiO =25.69%. To prepare a solution which would give a 10% nickel loading on the catalyst, 4.42grams of nickel nitrate was dissolved in 5ml distilled water.
3. Two grams of the dried catalyst was placed in 20ml beaker. Nickel nitrate solution in amount equal to the pore volume of the catalyst sample (0.88ml) was added slowly. The mixture was then stirred for 20 minutes to ensure proper mixing. It was then placed in an oven at 383K for 5 hours to evaporate water.
4. The sample was then placed in a quartz tube and calcined in a preheated tubular furnace at 823K for 8 hours under a slow flow of air (1-2mm/sec).
5. The sample was then cooled and stored for further experiment and characteri- zation. Deposition Mechanism 86 DepositionMehnim8 5.4 Characterization Results
Again, the characterization was conducted on the same dedicated scanning trans- mission electron microscopy VG HB5 with Link energy dispersive X-ray analyzer we have discussed in Chapter 4. The X-ray mapping and microanalysis were obtained at 100kev, with a nominal probe size of about 2nm.
5.4.1 Catalyst with Impregnated Nickel
The element mapping showed that the impregnated nickel and the catalytic metals, or Mo and Co, are all very well dispersed on the surface. A typical crystallite measures about 1.5 to 2nm, though that is the limit of the STEM instrument. Since no apparent crystallites could be observed within the limit of instrument. XEDS analysis was basically conducted on some random locations on the catalyst. The analysis result is shown in Figure 5-3. Apparently, nickel has no preference to either cobalt or molybdenum as expected.
5.4.2 Catalyst with Impregnated Nickel after Being Treated
The catalyst with nickel was then put into a reactor under the same condition with the hydrodemetallation experiment. The aim was to observe the development of the deposition on the catalyst surface. After about 200 hours under the temperature of 648K and hydrogen pressure of 4.8 MPa with 0.3% of hydrogen sulfide, the catalyst sample with impregnated nickel was characterized again by the STEM. Although some crystallites were observed on the surface, elemental mapping was not sufficient to provide the evidence for migration. On the other hand, the XEDS analysis result evidently indicates that nickel and coablt are associated with each other as shown in Figure 5-5. The normalized local loading of nickel deposits increases with the normalized local loadings of cobalt, which Deposition Mechanism 87 Deposition Mechanism 87 indicates the association between nickel and cobalt on the catalyst. Compared with Figure 4-7 and Figure 4-10, the association of cobalt with nickel is not as strong. It is due to either the shorter time, or the difficult of migration once the nickel deposits on the surface. It should also be pointed out that the average nickel composition in Figure 5-5 is slightly lower than the average composition on Figure 5-3. There are two possible explanations. There was some nickel loss during the processes, which caused the decreasing of nickel contents on the catalyst surface. Alternatively, it may simply a matter of sample selection. Figure 5-3 was obtained from analyzing 5 different catalyst particles, we do see there are some discrepancies between the particles. Figure 5-5 includes data from 6 catalyst particles.
5.5 Discussion
The following two questions can be raised with the experimental results on migration study:
* Why are the nickel deposits mobile?
* Why do the nickel deposits migrate towards cobalt?
These two questions will be discussed in the following. The first question is dis- cussed in two aspects: Tammann temperature and surface diffusivity. The question on the association between nickel and cobalt are discussed in terms of atomic com- patibilities between the metal sulfides.
5.5.1 Tammann Temperature
The higher the cohesive energy, the slower the migration is expected. Therefore, it is expected that the migration of metal or metal sulfides on a support surface should 88 Deposition Mechanism D M a 88
STEM
TEM Micrograph Molybdenum 8.13% I
- - - Nickel 10% I Cobalt 4.48%
Scale: 40nm
Catalyst: HDS16A Aging Time: 0 hrs.
H2 Pressure: 4.8MPa Temperature: K
Figure 5-2: Elemental Mapping of HDS16 with Impregnated Nickel Deposition Mechanism 89
A ^ V.>
0.8
o 0.7 0 :oo o~~~~ * O. 0 0 ,~~~~~~ .0 ':: o O O. O 0.6 ....D16 ypi Er wit. ys Of x ,10 *gnfd.Nj -r
. O O 0.5 IDS16A Catalyst with :10% Impxrgnated Nic~kel 0.4
%J. - 0.2 0.25 0.3 0.35 0.4 0.45 0.5 0.55 0.6 IIB 2 Co/(Co+Mo)
Figure 5-3: Microanalysis of HDS16A with Impregnated Nickel 90 DepositionDepsiio MechanismMehais 9
__ _ STEM TEM Micrograph Molybdenum 8.13%
Nickel 10% Cobalt 4.48%
.1 . , 1ti ,,N . ,-
Scale: 40nm
Catalyst: HDS16A Aging Time: 200 hrs. H2 Pressure: 4.8MPa Temperature: 648K
Figure 5-4: Elemental Mapping of HDS16 with Impregnated Nickel after Treating Deposition Mechanism 91
U.,.1 . _· __
0 HDS16A Catalyst with: 0.6 10%6:1iOregnd NickEI 648K, 2Wrs *0 Typical Error Bar o° , 0 0.5 . ~~.
Of 0 0 Oo b 0 0.4
, , .' .
O 0
0.3 0 0
0
0.2
N( I L I · I L I I 0.1lt5 0.2 0.25 0.3 0.35 0.4 0.45 0.5 0.55 0.6
Co/(Co+Mo)
Figure 5-5: Microanalysis of HDS16A with Impregnated Nickel after Treating Deposition Mechanism 92 Deposition Mechanism 92~~~~~~_ have a relationship with cohesive energy (and thus also with the melting point). The mobility of the atoms on the surface of a crystallite can significantly affect the migration process. Tammann temperature, TTam, which is defined as half of the melting temperature Tm of the bulk solid in degrees K, provides a measure of the extent of mobility of the atoms. Similarly, Hiittig temperature, one third of the melting point, is considered the temperature at which surface atoms become mobile. It is suggested that the Tammann temperature is associated with a two dimensional melting of the surface of the solid, i.e., to the transition from a solid to a liquid- like behavior of the surface[1ll]. Table 5.1 shows the melting points and Tammann temperatures of the relevant metal sulfides. The operating temperature range of 588- 623K is well over the Tammann temperature of nickel sulfide. As a consequences, we expect that the nickel deposits are relatively mobile on the catalyst surface and ready to migrate towards thermodynamically more stable form. The mobil ity of molybdenum sulfide and cobalt sulfide will be studied in chapter seven.
5.5.2 Surface Diffusivity
Assuming the sintering occurs because of the migration of the islands, effective diffu- sion coefficients of the order of 10-15 to 10- 17 cm 2 /sec were obtained[84][111]. Let's take a rough estimation of the diffusivity for the present work. The migra- tion distance on the catalyst surface is around 50nm, and the hydrodemetallation experiment ran typically about 500 hours. The effective diffusion coefficient can be estimated to be (r 2)/t, or 10- 17, which falls in the range of the literature data.
5.5.3 Affinity between Nickel and Cobalt Sulfides
The association between nickel and cobalt sulfides suggested that the formation of solid solution between the two sulfides. Next, we will discuss whether the conditions Deposition echanism 93 Dpi M 9
Table 5.1: Characteristic Temperatures of Metal Sulfides
Ni 7 S6 CogS8 MoS2
TMelting (°C) 790 1100 1750
TTammann (C) 258 414 738
THuttig (C) 81 185 401
Order of Mobility: Ni 7S6 > Co9S8 > MoS2
11 1..... Deposition Mechanism 94 DepsiioMehnim9 for the formation of solid solutions exist between the two metal sulfids. The conditions for formation of solid solutions are dependent upon the atomic compatibility between the two components. The principal relationship between phase diagrams and crystal chemistry is this: miscibility occurs when atoms have similar size, valence and structure, and compounds form when they do not[44][68].
* Ionic sizes;
* Valence;
* Molecular structure.
Field Strength
For metal oxides or metal sulfides, the major factors determining the extent of solid solutions are the relative ionic sizes and valences. Although different ionic sizes can definitely preclude extensive solid solution formation, valence difference can frequently be made up in other ways[44]. Using field strength as a parameter, Berkes & Roy [6] correlated several characteristics of binary phase diagrams of about 160 oxide systems. Field strength was defined as cation valence divided by the square of cation- anion distance (Z/d 2). The number of compounds in the binary system increases as a function of AL(Z/d2 ), the difference in the field strength of the end-member cations. As expected, the extent of solid solution is a maximum when A(Z/d 2 ) = 0 , and decreases rapidly as A(Z/d 2 ) increases. When A(Z/d 2 ) is less then 10%, extensive or complete solid solution take place. When A(Z/d 2 ) is large than 0.4, there are virtually no solid solution. In the present system, there exist three metal sulfides on the aged catalyst surface:
Ni 7S6 , Co9gS and MoS 2. The field strength can be easily calculated from the listed data. The results are listed in Table 5.2. The difference of field strength between Ni 7S6 and CosSs is Deposition Mechanism 95 Deoito Mehais ______95__
Table 5.2: Field Strength of Metal Sulfides
Cation Anion Cation Ionic Radii Ionic Radii Valence Field Strength Difference(%)
. _ ~(nm) (nm)
Ni 7S 6 0.069 0.170 +2 0.350
CogSs 0.072 0.170 +2 0.342 2.3%
MoS 2 0.079 0.170 +4 0.645 84.3%
References: [20] [37] [62] [81] [87]
· . , . . ,...... Deposition Mechaiiism 96 Deposition Mechanism 96 only about 2.3%, which is much less than 10%, extensive formation of solid solution between the two is expected. On the other hand, the difference between MoS 2 and the other two sulfides are significantly larger, therefore, little solid solution can be expected. Although no phase diagram between MoS 2 and Ni 7S6 is available, the phase diagram of Co-Ni-S showed in Figure 2-1 does show the extensive solid solution formation.
Molecular Structure
Another factor determining the extend of solid solution formation between two solids are their structures. Compared with the cubic structure of cobalt sulfide, molybdenum sulfide has a layered hexagonal structure, while nickel sulfide has an orthorhombic structure with structure parameters quite close to those of the cobalt sulfide, as can be seen in Table 5.3. Both the field strength and the molecular structure analyses indicate that exten- sive solid solution between Ni7 S6 and Co958 are expected. This is consistent with the phase diagram we showed in Chapter 2. On the other hand, solid solution formation between Ni 7 S6 and MoS 2 is not favored in any way.
5.6 Conclusions
Experiments were designed to differentiate two possible deposition mechanisms: ac- tivity or migration. Based on characterization of catalyst with impregnated nickel, in conjunction with data in the literature on activity study, the following two questions were answered in this chapter concerning the development of the nickel deposits on the catalyst surface. First of all, the nickel deposits are mobile on the catalyst surface due to the fact that the hydrodemetallation temperature is higher that the Tammann temperature of nickel sulfide. Secondly, the association between nickel and cobalt is Deposition Mechanism 97
Table 5.3: Structures and Properties of Metal Sulfides
a M.W. Space Group b alpha Note C
MoS 2 161.14 Fm3m 0.31602 Hexagonal 1.2294
CosS8 786.88 P63/mmc 0.99273 Cubic
Ni 7S 6 603.19 Cmcm 0.918 close to cubic Orthorhombic 1.1263 0.9457 . - - . , . I~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Deposition Mechanism 98
resulted from the formation of solid solution between the two sulfides. Although we do not have experimental basis to claim that nickel initially deposits on molybdenum sites, it is a reasonable postulation since the activity of molybdenum sulfide for hydrodemetallation. Further experiments are necessary to conclusively determine the initial distribution of nickel on the catalyst surface. Chapter 6
Metal Distribution within Deposition Crystallites
It may, of course, be trivial -- individual eccentricity; or it may be very much deeper than appear on the surface. -Sherlock Holmes, The Adventure of the Red Circle Sir Arthur Conan Doyle
6.1 Chapter Summary
Structure of nickel deposits on the catalyst surface and the interaction between the de- posits and the catalytic metals were characterized by scanning transmission electron microscopy (STEM) and high resolution transmission electron microscopy (HRTEM) analyses. The deposits were in crystallite form of nickel sulfide on the catalyst sur- face. Within the crystallites, it is found that cobalt sulfide is uniformly distributed
99 Metal Distribution within Deposition Crystallites 100
throughout the crystallites. In contrast, only about 20% to 25% of the molybdenum sulfide is associated with the nickel sulfide deposits as a segregated surface layer. The rest of the moybdenum sulfide forms seperate entities on the substrate. For crys- tallites smaller then about 15 nanometers, the degree of segregation decreases. The implications on the deactivation of hydrodemetallation catalysts are discussed.
6.2 Introduction
In the previous two chapters, resorting to X-ray elemental mapping, we showed that nickel deposits are strongly associated with cobalt on the catalyst surface, but not with molybdenum. Since molybdenum sulfide on sulfided CoMo/ - A1203 is gen- erally considered to be the active component for hydrotreating reactions[16][23], it is important to know how the nickel deposits interact with molybdenum on the surface. The objective of this work is to further characterize the structure of the deposits and the interaction between the deposits with catalytic metals, especially the distribution of molybdenum on the aged catalyst surface, and further the elemental distributions within the nickel deposit crystallites.
6.3 Characterizations
6.3.1 Characterization by STEM
As we have shown in previous chapters, nickel sulfide deposits are in crystallite forms on the catalyst surfaces. The sizes of the crystallites were estimated to be about 10-15 nanometers, though crystallites with sizes up to 50 nanometers could be observed. By positioning the STEM electron probe on some crystallites and conducting EDS analysis, we concluded that nickel is strongly associated with cobalt on the catalyst. However, the structure of molybdenum and its relation with the nickel deposits were Metal Distribution within Deposition Crystallites 101 _ _
not discussed. Element distributions within the crystallites are studied in this chapter. The aim is to detect element distributions within the crystallite, more specifically, the existence of a surface enrichment of one component. The idea is schematically shown in Figure 6-1. If the elements are uniformly distributed within the crystallites, we should get the same element ratios when we analyze the center or the edge of the crystallites. On the other hand, if there is an element enriched on the surface of the crystallites, we would get different element ratios at different analyzing positions. Unfortunately, the feature of problems and the limitation of the instrument determine that we could only get qualitative results. The first sample we analyzed was an aged catalyst HDS16A sample, which is loaded with 23%wt nickel. To '.cilitate the analysis, we get on to magnification of x 1,000,000, and obtained the best resolution we could. At this condition, we moved the sample and located some large crystallites. Figure 6-2 is a TEM micrograph of the area where the crystallite with square shape at the center would be analyzed. The crystallite measures about 50 nanometers. First, we focused the electron probe on the very edge of crystallite and conducting XEDS analysis, which continued for 100 seconds. During the analysis, we would check the sample drifting every 20 seconds to make sure the probe was still on the original positions. After the analysis is complete, we move the probe onto another position. The analysis results are plotted in Figure 6-3 for the 15 positions analyzed on the particular crystallite. The circles indicate the positions analyzed, and the num- bers inside are the relative amounts of cobalt or molybdenum to local nickel con- tents. Clearly, there is a significant difference between the distributions of cobalt and molybdenum. Although cobalt is virtually uniform throughout the whole crystallite, molybdenum is strongly enriched on certain surfaces of the crystallite. The result is also plotted in Figure 6-4 as the element ratio to the relative distance from the center Metal Distribution within Deposition Crystallites 102 ______
_ _ ___
Electron Beam
Component A Segregated
Components A and B Uniformly Distributed
Figure 6-1: Illustration of the Electron Probe on a Crystallite Metal Distribution within Deposition- -- Crystallites 103
--
I I
Scale: 50nm
Catalyst: HDS16A Temperature: 623K
Aging Time: 650hrs. H2 Pressure: 4.8 MPa
Figure 6-2: Electron Micrograph of a Crystallite Analyzed Metal Distribution within Deposition Crystallites 104 of the crystallite. The error bars represent standard deviations for multiple analyses at the same relative radial positions. The overall ratio of cobalt to nickel within the crystallite was analyzed as 0.17. Considering the small area analyzed, it is very close to the bulk ratio of 0.22. On the other hand, the molybdenum to nickel ratio is much lower than the bulk ratio of about 0.4. These results indicate that much of cobalt on the catalyst is associated with nickel, while only part of the molybdenum is asso- ciated with the nickel deposits as a segregated surface layer of molybdenum sulfide. The rest of the molybdenum sulfide forms seperate entities on the substrate. We also noted that molybdenum enrichment only occurred at certain particular surfaces, rather than every surface of the nickel deposits. Another aged catalyst SN6931 is also characterized with the similar procedure. As shown in Table 3.2, this is a catalyst with lower cobalt and molybdenum loadings. The nickel loading was about 22.1%wt. Figure 6-5 and Figure 6-6 show the micrograph and the analysis results for one crystallite with the size of about 50 nanometer. We can see the results are very similar. Cobalt us distributed throughout the crystallite, while molybdenum has a much higher concentration on the surface. The result is also plotted in Figure 6-7. Again, the ratio of cobalt to nickel is close to the bulk ratio of 0.13, while the molybdenum to nickel ratio much lower than the bulk ratio of 0.2. To make sure the analyzed crystallites are representative of the nickel deposits, and the results represent at least a qualitative trend, we analyzed more crystallites in the catalyst HDS16A with different sizes ranging from about 6nm to 60 nm. The sketches of some of the crystallites are shown in Figure 6-8. The results are plotted in Figure 6-9 and Figure 6-10. Once again, the ratios of cobalt to nickel are close to the bulk ratio of 0.2, while the molybdenum to nickel ratio of about 0.1 only accounts for about 25% of the bulk molybdenum to nickel ratio. Another interesting observation was made for the crystallites with different sizes. Similar to two component system, we define the following ratio as the surface enrich- Metal Distribution within Deposition Crystallites 105
Molybdenum (relative to 100 nickel)
Cobalt (relative to 100 nickel)
______
Figure 6-3: Element Ratios within the Crystallite Analyzed Metal Distribution within Deposition Crystallites 106
0.30
0.204 F
4 I
0 0.15
I I d Co/Ni I I 0.10 O Mo/Ni
U
0.05
I I I I I I - I - 0.00 0.25 0.50 0.75 1.00
Relative Distances from Crystallite Center, r/R
Figure 6-4: Element Distribution within the Crystallite Analyzed Metal Distribution within''' Deposition Crystallites-" 107
I I ur
I.-q
I·
Scale: 10nm
Catalyst: SN6931 Temperature: 588K Aging Time: 380hrs. H2 Pressure: 4.8 MPa
Figure 6-5: Electron Micrograph of a Crystallite Analyzed Metal Distribution within Deposition Crystallites 108
______
Molybdenum (relative to 100 nickel)
Cobalt (relative to 100 nickel)
______
Figure 6-6: Element Ratios within the Crystallite Analyzed ______Metal Distribution within Deposition Crystallites 109
0.125
0.100
I
i 0.075 I
S I IZoos0.0 I 0 QII
0.025
0
1 1 ~ ~ ~~I ~I ~~i I I (J I I I I I I ~~~~~I I 0.00 0.25 0.50 0.75 1.00
Relative Distances from Crystallite Center, r/R
Figure 6-7: Element Distribution within the Crystallite Analyzed Metal Distribution within Deposition Crystallites 110 _____
Figure 6-8: Sketches of Seven Crystallites Analyzed Metal Distribution within Deposition Crystallites 111 ______I __ __ ___
0.5
0.4
0.3
0.2
0.1
0.0 0.0 0.2 0.4 0.6 0.8 1.0 r/R
Figure 6-9: Molybdenum/Nickel Radial Distribution within Crystallites Metal Distribution within Deposition Crystallites 112 ______
0.4
0.3
' 0.2
0.1
0.0 0.0 0.2 0.4 0.6 0.8 1.0 r/R
Figure 6-10: Cobalt/Nickel Radial Distribution within Crystallites Metal Distribution within Deposition Crystallites 113
ment factor as in a binary system:
Xtrsrface Xyenter _AA X =x( wysurface)ycenter) XB rgB
where A represents the catalytic metals cobalt or molybdenum, and B the nickel deposits. When we plot the surface enrichment factor against the sizes of the crystallites in Figure 6-11, it is clear that cobalt is uniformly distributed with the crystallites regardless of the sizes of the crystallites, indicated by the fact that the enrichment factor is distributed evenly around one. Molybdenum, however, has a much higher concentration on the surface, though the enrichment factor decreases for crystallites smaller than about 15 nanometers. It is consistent with the theoretical calculations by Helms[33]. By applying a simple mass balance relationship to segregation equation, Helms presented a plot of segregation vs the dispersion of particles, which is defined as the ratio of the number of atoms on the surface to the number of the total atoms. It was noted that the effect of particle size could be very significant, depending on the magnitude of the heat of segregation {-AG/RT}. Intuitively, as the particle size approaches zero, the surface composition must approach the bulk value. For crystallites smaller than about 5 nanometers, we were limited by the resolution of the scanning transmission electron microscope. In addition, the drifting of the samples also causes difficulties for the analysis of even smaller particles.
6.3.2 Characterization by HRTEM
Although the structure of sulfided CoO - MoO3/y - Al 2 03 catalyst has been studied extensively by high resolution electron microscopy[18][20] [21][80], HRTEM studies for aged catalyst are still scarce[108] [116]. In the last section, we analyzed aged catalyst samples by scanning transmission Metal Distribution within Deposition Crystallites 114
2.0
1.6
1.2
00.8
-- q
0.4
nn 0 10 20 30 40 50 60
Size of Crystallites, (nm)
Figure 6-11: Effect of Crystallite Sizes on Segregation Metal Distribution within Deposition Crystallites 115
electron microscope, and concluded that there was a segregated surface layer of molyb- denuml disulfide on the surface of the nickel sulfide deposits. In the following section, we will present results from high resolution transmission electron microscope to di- rectly show the structure of the deposits, and the surface phase of molybdenum sulfide. Figure 6-12 and 6-13 show two areas of a typical TEM micrograph of the aged HDS16A catalyst sample. Typically, we could see many crystallites on the cata- lyst surfaces, which STEM analysis and XRD confirmed to be Ni 7 S6. Many dif- ferent lattice fringe images car be observed on the micrographs. The 111 reflection
(d=0.575nm) of Ni7S 6 can be observed within many crystallites(Figure 6-12). The lattice fringes observed in the figure have spacings of about 0.62nm that one can easily relate to the 0.615nm spacings of the 002 basal planes of MoS 2. The structure made of highly disordered S-Mo-S layers of poorly crystallized MoS 2. Due to lattice relaxation, some of the fringe spacing are larger than 0.615nm. Consistent with the results from XEDS analyses, molybdenum sulfide was ob- served on the surfaces of many of the nickel sulfide crystallites. While most of the nickel sulfides have one or two layers of molybdenum sulfide on the surfaces, crystal- lites with as many as five or six layers of molybdenum sulfide was also observed, as shown in Figure 6-13. The fringe spacing of 0.33nm corresponds to the 131 reflection
(d=0.328nm) of Ni 7S 6. For slabs of more than two layers, it is easy to assign them to the basal planes of MoS 2. It is, however, difficult to determine its nature, when there exists only one layer, as in many of the crystallites. According to the shape of the lines, and information from STEM, we believe the thick line around many of the
Ni 7 S6 crystallites are a single layer of MoS 2 slab. Figure 6-14 shows a molybdenum sulfide crystallite. The lattice fringe spacing of 0.28 nm corresponds to the 100 reflection (d=0.274nm). The Moire fringe is probably caused by overlapping of two crystals. The hexagonal shape of the crystal is very well Metal Distribution within Deposition Crystallites 116
Scale: 5nm
Catalyst: HDS16A Temperature: 623K
Aging Time: 650hrs. H2 Pressure: 4.8 MPa
Figure 6-12: Lattice Fringe Image of Aged HDS16 Catalyst Metal Distribution within Deposition Crystallites 117 _ __
Figure 6-13: Lattice Fringe Image of Aged HDS16 Catalyst Metal Distribution within Deposition Crystallites 118 defined. The size of the crystal is about 30 nanometers. Although it is a rarity, the size of the crystal is very significant. No lattice fringes are definately assigned to CogSs, partly due to the fact that the main reflections 111 (d=0.573nm) and 002 (d=0.496nm) of Co09S are difficult to be differentiated with those reflections of Ni7 S6 (dlll=0.575nm, doo2=0.470nm). The difficulty of locating lattice fringes of Co0gS could also be explained by the formation of solid solution between Co095 and Ni7 S6. Many fringe images with spacing at 0.15 to 0.2 can be observed on the pictures, though it is difficult to assign them to a particular compound.
6.3.3 Characterization by XPS
X-ray photoelectron spectroscopy was used characterized the aged catalyst surface. Although most of the XPS literature work has been addressing the chemical state of elements on the catalysts[12], we attempted to get a depth profile of elements with the crystallites. The original aim was to quantify the average thickness of the molybdenum sulfide surface phase on the nickel deposits crystallites by sputtering the catalyst surfaces. However, the actual catalyst surface is far more too complete than the XPS is designed for. First of all, the catalyst is a flat surface. Secondly, the the thickness of the molybdenum sulfide is probably beyond the resolution of the instrument. In addition, the binding energy shifts and peak broading caused by charging effects also make chemical state assignments uncertain. In retrospect, we now understand that most of the molybdenum is still on the substrate, rather than the surface of nickel sulfide. Consequently, it is probably impossible to get an average depth profile as we had expected. Metal Distribution within Deposition Crystallites 119 ______
Scale: 5nm
Catalyst: HDS16A Temperature: 623K
Aging Time: 650hrs. H2 Pressure: 4.8 MPa
Figure 6-14: Electron Micrograph of a Molybdenum Sulfide Crystallite Metal Distribution within Deposition Crystallites 120 I _ 6.4 Discussions
Surface segregation in metallic alloys[14] [41] and metal oxides [72] has been studied extensively in material science. There are several possibilities for the microstructure of such systems. One phase or one component could be enriched on the surface of the crystallites. At the present time, a comprehensive theory does not exist, but the following general conclusions have been proposed[8] [32] [97]:
* In one phase alloys the surface tends to be enriched by the component with the lower surface energy;
* When an alloy contains two phases in equilibrium, the alloy with the lower sublimation energy tends to form the outer surface;
* The degree of enrichment decreases with the increasing temperature.
Co-Ni alloy has been studied by using XPS, SIMS and other techniques, and the two metals form a continuous series of solid solutions close to ideal ones[65]. CoO-NiO system has also been studied[11] [66]. They are very well soluble in each other. In the present system, Co958 has a cubic structure. Although Ni 7S6 has an orthorhombic structure, it is very close to cubic. The lattice constants of the two are also very close to each other. The difference between the respective constants are about 10%. It is expected that these two sulfides can form solid solutions.
MoS 2 is one of the transition metal dichalcogenides, which belong to a large class of the so-called two-dimensional solids (Figure 6-15). Of course, we are actually deal- ing with three dimensional solids with strong anisotropy in their physical properties. They are called two dimensional because they are formed in layered structures. Atoms within a layer are bound by strong covalent or ionic forces while individual layers are held together by relatively much weaker forces. The latter are frequently referred Metal Distribution within Deposition Crystallites 121
to as 'van der Waals' type of interactions. Among all the transition metal dichalco- genides, MoS 2 has the highest degree of anisotropy on the basis of force constants determined from phonon studies[52]. Due to the special layered structure of MoS 2, it can be very readily cleaved along the basal plane. From a quasichemical point of view, surface energy of solid is determined by the total bonding energy involved when the surface is formed. Therefore, we know that
MoS 2 would have a much lower surface energy when being cleaved along the basal plane. The other two sulfides involved in the present system have cubic or othorhombic structures. These are no special cleavage planes for such structures. Thus, when the three solids are present in a system, molybdenum sulfide would be expected to be segregated onto the surface to achieve the minimum of energy of the total system. In addition to the surface energy differences between the two metal sulfides, an- other energy contribution involved in the segregation of molybdenum sulfide is the interaction energy at the interface. Naturally, the interaction would be different for different planes. This is shown by the fact that molybdenum sulfide only segregated on some surfaces of nickel sulfide crystals, rather than every surface. Summarizing the electron microscopic observation and the discussion, we can roughly portray the aged catalyst surface with the following physical picture. The metal sulfides on the catalyst surface form three different entities. Most of the nickel sulfide deposits are associated with cobalt sulfides, forming uniform crystallites. For most of the nickel-cobalt sulfide crystallites, a surface layer of molybdenum was ob- served. This part of molybdenum accounts for about 25% of the total molybdenum on the catalyst surface. The rest of the molybdenum is not directly associated with the nickel deposits, although this part of the molybdenum sulfide is still well dispersed, crystallites with sizes up to 30 nanometers were observed. Metal Distribution within Deposition Crystallites 122 __
A 'A kP